Вы находитесь на странице: 1из 377

Kyoji Sassa

Hiroshi Fukuoka
Fawu Wang
Gonghui Wang
Progress in Landslide Science
Kyoji Sassa
Hiroshi Fukuoka
Fawu Wang
Gonghui Wang

in Landslide Science

With 431 Images, 349 in Color


Sassa, Kyoji
President of the International Consortium on Landslides
Research Centre on Landslides, Disaster Prevention Research Institute,
Kyoto University, Uji, Kyoto 611-0011, Japan
Tel: +81-774-38-4110, Fax: +81-774-32-5597, E-mail: sassa@SCL.kyoto-u.ac.jp

Fukuoka, Hiroshi
Research Centre on Landslides, Disaster Prevention Research Institute,
Kyoto University, Uji, Kyoto 611-0011, Japan
Tel: +81-774-38-4111, Fax: +81-774-38-4300, E-mail: fukuoka@SCL.kyoto-u.ac.jp

Wang, Fawu
Research Centre on Landslides, Disaster Prevention Research Institute,
Kyoto University, Uji, Kyoto 611-0011, Japan
Tel: +81-774-38-4114, Fax: +81-774-38-4300, E-mail: wangfw@landslide.dpri.kyoto-u.ac.jp

Wang, Gonghui
Research Centre on Landslides, Disaster Prevention Research Institute,
Kyoto University, Uji, Kyoto 611-0011, Japan
Tel: +81-774-38-4114, Fax: +81-774-38-4300, E-mail: wanggh@landslide.dpri.kyoto-u.ac.jp

Library of Congress Control Number: 2007920433

ISBN-13 978-3-540-70964-0 Springer Berlin Heidelberg New York

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitations, broad-
casting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this
publication or parts thereof is permitted only under the provisions of the German Copyright Law of
September 9, 1965, in its current version, and permission for use must always be obtained from
Springer. Violations are liable to prosecution under the German Copyright Law.

Springer is a part of Springer Science+Business Media

Springer-Verlag Berlin Heidelberg 2007
Printed in Germany

The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.

Cover Photo: Guinsaugon landslide, Southern Leyte, Philippines

Cover design: deblik, Berlin

Typesetting: Stasch Bayreuth (stasch@stasch.com)
Production: Almas Schimmel

Printed on acid-free paper 32/3141/as 5 4 3 2 1 0

Foreword for the Publication of
Progress in Landslide Science from UNESCO

Natural disasters are increasing in terms of frequency, complexity, scope and destruc-
tive capacity. They have been particularly severe during the last few years when the
world has experienced several large-scale natural disasters: the Indian Ocean earth-
quake and tsunami, Hurricane Katrina and Hurricane Rita; the Kashmir earthquake
in Pakistan; floods and forest fires in Europe, India and China; and drought in Africa.
Images of these events have shocked us all and will remain with us for a long time.
Numerous landslides and mudflows have also occurred, causing deaths, injuries and
material losses. The most recent tragic ones were the large-scale landslides which struck
the Philippines in 2006, hitting the Albay province on 02 December and the Leyte Is-
land on 17 February respectively, resulting in terrible loss of life, suffering and dam-
age. National authorities and the international community, of course, should continue
to provide the practical support needed by the affected communities. At the same
time, it is important to quickly learn appropriate lessons that may help individuals,
families, communities and whole societies to be better prepared for other disasters,
whether caused by natural forces or otherwise.
The time has come for putting more emphasis on pre-disaster action rather than
remaining content with post-disaster reaction. We must mobilize scientific knowl-
edge and technological know-how to assess natural hazards and to strengthen disas-
ter mitigation measures. We should promote a better understanding of natural disas-
ters. We must promote and enforce sound scientific, engineering and construction
principles. And we must promote education and public awareness about natural di-
saster reduction.
Landslides pose considerable risks to peoples livelihoods and to the environment.
They cause great disruption and economic losses by the destruction of infrastructure
works such as roads and other communications and utility lines and of cultural heri-
tage and the environment. Today there is a need more than ever before to address the
problem of landslides in an integrated and internationally concerted way. These are
the purposes of the International Consortium on Landslides (ICL) and the Interna-
tional Programme on Landslides (IPL). Both initiatives encompass research, educa-
tion and capacity-building in landslide risk reduction. They both enjoy the participa-
tion and support of numerous international, governmental and non-governmental
organizations and entities. They contribute to the International Strategy for Disaster
Reduction (ISDR) and to the implementation of the Hyogo Framework for Action
20052015 which was adopted at the World Conference on Disaster Reduction held in
Kobe, Japan in January 2005. The 2006 Tokyo Action Plan on Landslides, which was
adopted during the Tokyo Round Table Discussion on Landslides in January 2006,
provides a roadmap for strengthening international collaboration and identifying fo-
cus areas for reducing landslide risk worldwide.
UNESCO had the privilege to accompany from the very beginning the establish-
ment of the ICL and the launching of the IPL. In so doing, the Organization enjoys
partnership with a large number of stakeholders including the World Meteorological
Organization (WMO), the United Nations University (UNU), the Food and Agricul-
ture Organization (FAO) of the United Nations, the UN/ISDR Secretariat and the In-
VI Foreword for the Publication of Progress in Landslide Science from UNESCO

ternational Council for Science (ICSU) and its Unions. I am glad that the ICL and IPL
have also been marked with the establishment of a UNITWIN Cooperation Pro-
gramme on Landslide Risk Mitigation for Society and the Environment in the frame-
work of the UNITWIN/UNESCO Chairs Programme, at Kyoto University (KU). This
Programme is now hosted in the UNESCO-KU-ICL UNITWIN Headquarters at the
Research Centre on Landslides of the Disaster Prevention Research Institute in Kyoto
University. Furthermore UNESCO and ICL have established in August 2006 a Memo-
randum of Understanding for cooperation. Finally I am especially pleased that
UNESCO will serve as co-organizer with ICL of the World Landslide Forum sched-
uled to take place in 2008. This Forum will constitute a milestone in our efforts to
strengthen global risk preparedness.
The Landslides Journal of the International Consortium on Landslides plays a
key role for the progress of landslide study as an integrated research field by putting
together knowledge and technologies in many related fields of natural sciences, engi-
neering, social sciences and culture. The present Publication Progress in Landslide
Science comes in this context to provide an overview of the current status of this
science. The diversity of subjects which are presented in this publication represents a
rich collaborative work regarding landslides. I wish to commend the editors and the
numerous authors involved in it. My particular greetings go to Professor Kyoji Sassa,
Chairperson of ICL who continues to spare no effort in promoting ICL and IPL. It is
with great pleasure that I praise the edition of this publication as a means of dissemi-
nating good knowledge in the area of landslide risk reduction.

Kochiro Matsuura
Director-General of UNESCO
Foreword for the Publication of
Progress in Landslide Science from UN/ISDR

A series of extremely high-profile disasters the Indian Ocean tsunami of December

in 2004, Atlantic hurricane season, the South Asian earthquake and the East African
drought in 2005 underscored the importance of how better cooperation between Gov-
ernment authorities and the international community including scientific commu-
nity would have played a critical role in helping people make life changing decisions
about where and how they live before the disaster strikes, in particular high-risk
urban areas.
Landslide, floods, drought, wildfire, storms, tsunami, earthquakes and other types
of natural hazards are increasingly affecting the world. In the decade 19761985, close
to billion people were affected by disasters. But by the most recent decade, 19962005,
the decade total had more than doubled, to nearly two and a half billion people. In the
last decade alone, disasters affected 3 billion people, killed over 750 000 people and
cost around US$ 600 billion1. We cannot let this trend continue. Disaster risk concerns
every person, every community, and every nation; indeed, disaster impacts are slow-
ing down development, and their impact and actions in one region can have an impact
on risks in another, and vice versa. Without taking into consideration the urgent need
to reduce risk and vulnerability, the world simply cannot hope to move forward in its
quest for sustainable development and reduction of poverty.
The Hyogo Framework for Action 20052015: Building the Resilience of Nations and
Communities to Disasters, adopted at the World Conference on Disaster Reduction
(WCDR, Kobe, Hyogo, Japan, in January 2005), represents the most comprehensive
action-oriented policy guidance in universal understanding of disasters induced by
vulnerability to natural hazards and reflects a solid commitment to implementation
of an effective disaster reduction agenda. In order to ensure effective implementation
of the Hyogo Framework at all levels, tangible activities must be carried out. For the
last two years as post WCDR, we have seen many activities and initiatives developed to
implement the Hyogo Framework in various areas. As a concrete activity in the area of
landslide risk reduction, the International Programme on Landslides has maintained
the momentum created at the WCDR and has been moving forward, led by the Inter-
national Consortium on Landslides.
This publication is a valuable contribution to the implementation of the priority
area 2 of the HFA Identify, assess and monitor disaster risks and enhance early
warning, by gearing landslide risk assessment, both hazard identification, monitor-
ing and vulnerability analysis, as well as preparedness and landslide risk manage-
ment. The combination of landslide scientific knowledge and risk reduction measures
are essential to reduce the impact of landslides. The Hyogo Framework calls for the
international coordination and collaboration among different actors dealing with di-
saster risk reduction. In this sense, the Global Cooperation Platform for research and
investigation for landslide risk reduction in the 2006 Tokyo Action Plan consist of very

Data derived from the EM-DAT: The OFDA/CRED International Disaster Database, www.en-dat.net,
Universite Catholiqu de Louvain, Brussels, Belgium.
VIII Foreword for the Publication of Progress in Landslide Science from UN/ISDR

important activities to promote the thematic coordination to research and reduce

impacts of landslides. This initiative also contributes to the priority area 3 of the HFA,
which emphasizes the importance of education and public awareness of the disaster
risk reduction. Education and public awareness about the hazards, in this case land-
slide, are also key for people to be able to reduce risks and their vulnerabilities.
I welcome the work of the Global Cooperation Platform for research and investi-
gation for landslide risk reduction and the International Consortium on Landslides.
I look forward to collaboration with them, in particular through the International
Programme on Landslides (IPL) and its Global Promotion Committee.

Slvano Briceo
Director of United Nations Secretariat
of the International Strategy for Disaster Reduction
UNESCOs Contribution to Landslide Risk Reduction

International scientific programmes provide a forum for an in-depth study of natural

phenomena, their characteristics and their occurrence. This study is an essential pre-
requisite for a logical approach to the understanding of natural hazards. Furthermore
the mitigation of risks arising from these hazards is based on the applied science and
technology as well as on educational and information campaigns and programmes.
The problem of risk reduction therefore requires a multidisciplinary approach, in-
volving co-operation between specialists in several sectors of science, technology, edu-
cation and culture.
The activities of the United Nations Educational, Scientific and Cultural Organiza-
tion (UNESCO) in the study of natural disasters and the protection against them date
from the beginning of the 1960s. Originally concerned with basic seismology, these
activities were later extended to the reduction of earthquake hazards and still later to
other categories of natural hazards including landslides and their socio-economic as-
pects. UNESCO has brought an interdisciplinary approach to the study of geohazards
and the mitigation of their effects. Being at the crossroads of several sectors, UNESCO
provides a unique intellectual setting, linking, within a single organization, the natu-
ral sciences with education, culture, communication and the social sciences. With its
broad mandate and breadth of expertise, UNESCO is able to integrate many of the
essential ingredients for disaster reduction.
The purposes of UNESCO in the field of disaster risk prevention can be described
as follows:

 to promote a better understanding of the distribution in time and space of natural

hazards and of their intensity;
 to help set up reliable early warning systems;
 to foster rational land use plans;
 to encourage the adoption of suitable building design;
 to provide policy advise on the protection of educational buildings and cultural
 to strengthen environmental protection for the prevention of natural disasters;
 to enhance preparedness and public awareness through education and training;
 and, when catastrophes do strike, to foster post-disaster investigation, recovery and

Facing the increasing vulnerability of its Member States to natural hazards, UNESCO
has constantly advocated that risk prevention policies, including warning systems re-
lated to natural hazards like landslides, must be established or improved. Hence the
promotion of landslide risk reduction, environmental protection, and sustainable de-
velopment has become among the objectives of UNESCO. Various studies have been
conducted, and efforts supported by the Organization on the cause and prevention of
landslides. In the 1970s, UNESCO books including guidelines on landslides hazards
zonation have been published. As early as 1981 a joint project initiated jointly with the
United Nations Environment Programme (UNEP), under the title The Protection of
X UNESCOs Contribution to Landslide Risk Reduction

the Lithosphere as a Component of the Environment, resulted in significant research

being carried out on landslide mitigation. Both the International Hydrological Pro-
gramme (IHP) and the International Geoscience Programme (IGCP) have taken promi-
nent roles in promoting activities related to landslides. The International Flood Ini-
tiative, which is led by UNESCO, addresses landslide risks as an integral part of hydro-
logical extremes.
The foundation of the International Consortium on Landslides (ICL), in itself, lies
within an IGCP project titled Landslide Hazard Assessment and Mitigation for Cul-
tural Heritage Sites and other Locations of High Societal Value. Approved in 1998,
this project served as an initial platform for the establishment of a Memorandum of
Understanding concerning cooperation in research for landslide risk mitigation and
protection of the cultural and natural heritage between UNESCO and the Disaster
Prevention Research Institute, Kyoto University, Japan.
UNESCO has been closely associated with the process that led to the establishment
of ICL in January 2002. Subsequently the UNESCO UNITWIN Programme on Land-
slide Risk Mitigation for Society and the Environment came into effect in 2003, fol-
lowed by the construction in 2004 of the headquarters building of the UNESCO/Kyoto
University/ICL UNITWIN cooperation program within the Research Centre on Land-
slides, which hosts the secretariat of the International Programme on Landslides (IPL).
Under the above-mentioned cooperative mechanisms, a number of undertakings
have been carried out, including the landslide investigation in the Incas World Heri-
tage Machu Picchu in Peru. Above all, UNESCO has fully participated in the adoption
of the 2006 Tokyo Action Plan on Landslides which represents a global road map for
the assessment and mitigation of landslides.
Much more recently an understanding has been reached between UNESCO and
ICL in order to cooperate in the following areas:

1. The promotion of landslide research for the benefit of society and the environ-
ment, learning and capacity-building in landslide risk reduction, notably in devel-
oping countries;
2. the integration of earth sciences, water sciences, geophysical and geotechnical sci-
ences, technology and disaster management within the appropriate cultural and
social contexts in order to evaluate landslide risk in urban and rural areas, includ-
ing cultural and natural heritage sites, as well as to contribute to the protection of
the human and natural environment, including lifelines and buildings of high soci-
etal value;
3. the promotion of some or all of the global cooperation fields that have been agreed
in 2006 Tokyo Action Plan.

UNESCO expects to enhance its contribution to the above-mentioned areas in shap-

ing its participation in the preparations and follow-up of the First World Landslide
Forum to take place in Tokyo from 18 to 21 November 2008. This participation will
build on interdisciplinary scientific work through and among UNESCOs intergovern-
mental scientific programmes, and will capitalize further on the transdisciplinary ac-
tivities between the programmes of the Organization in the sciences, education, cul-
ture and communication.

Badaoui Rouhban
Section for Disaster Reduction
Natural Sciences Sector
Establishment of the Technical Journal
Landslides as the Successor to Landslide News

Landslides pose a hazard and threat in most countries. Worldwide, they annually cause
billions of dollars in damages and thousands of casualties. To counter these losses, scien-
tists and engineers are successfully conducting research that aims to investigate and miti-
gate landslide hazards. To sustain this research requires continuing communication among
researchers and those who apply its results.
From 19872003 this goal of communication between researchers and practitio-
ners was aided by the international newsletter Landslide News, which was edited,
published, and distributed by the Japan Landslide Society. Professor Sassa served as
Editor-in-Chief of this comprehensive newsletter, which was a very successful techni-
cal news publication. It reported on landslide news, landslide research, the efforts of
landslide research organizations, and news of landslide-related technical meetings.
At the first session of the Board of Representatives of the International Consortium on
Landslides (ICL), which was held at UNESCO Headquarters in Paris on 1921 November
2002, the Board decided to launch the International Programme on Landslides (IPL). As
the first coordinating project of this new program, the Board conceived the concept of the
new scientific journal Landslides as the successor to Landslide News. This new jour-
nal was planned and activated by ICL as the international organ of scientific communica-
tion on landslides. Thereafter, extensive deliberation on the necessary financial resources,
editorial board, editorial secretariat, and publishing firm was conducted. As a result of
this deliberation, Professor Sassa was appointed Editor-in-Chief of this new journal and
Kyoto University was named as the site of the secretariat.
The first issue of Landslides was published by Springer-Verlag in April 2004 as the
successor to Landslide News. This new quarterly journal with color illustrations was
created by and for landslide researchers and practitioners as a means of communication
to promote landslide research and to contribute to landslide disaster reduction.
Landslides is supported by the United Nations Educational, Scientific and Cultural
Organization (UNESCO), the World Meteorological Organization (WMO), the Japan Min-
istry of Education, Culture, Sports, Science and Technology (MEXT) and Kyoto Univer-
sity. In addition to these organizations, ICL-supporting organizations and landslide ex-
perts worldwide are cooperating in writing, editing, and publishing this journal.
Publication of Landslides represents an important landmark in promoting re-
search on landslide hazards worldwide, and thus in the mitigation of landslide haz-
ards. The journal contributes to the development and continuance of landslide re-
search by providing an international forum for the exchange and coordination of ex-
pertise in landslide risk assessment and mitigation.
We thank Professor Sassa for his efforts as Editor-in-Chief of Landslide News and
its successor, the very successful technical journal Landslides. We wish this new jour-
nal a long life in service of the international landslide research community.

Robert L. Schuster
U.S. Geological Survey, E-mail: rschuster@usgs.gov
Preface Aims of This Volume

Large and small landslides occur almost every year in nearly all regions of the world.
However, the number of landslides is difficult to ascertain, and even the number of
landslide-caused casualties is not correctly counted worldwide. Most casualties caused
by rain-induced landslides are included in those tabulated for hurricane or storm di-
sasters, and casualties caused by earthquake-induced landslides are often included in
those for earthquake disasters. Thus, the casualties due to landslide disasters are often
extremely underestimated. Japan has statistics of casualties by various types of land-
slides (small and large debris or rock slides, debris flows, rock falls, et al. since 1967),
even though they occurred during typhoons, earthquakes, or volcanic activities. Fig-
ure 0.1 shows the statistics of casualties caused by landslides, earthquakes, and volca-
nic activities in Japan for the period of 19672004. Casualties by earthquake-induced
landslides are included both in the landslide disasters and earthquake disasters. Land-
slide disasters in Japan for this period have occurred every year; the total number of
deaths (3 285) due to landslides is about one half of the deaths (7 008) caused by earth-
quakes, including the catastrophic 1995 Kobe earthquake. Extensive landslide preven-
tion works have been constructed in Japan. Thanks to those preventive works, the
number of casualties has gradually decreased as seen in Fig. 0.1 in spite of progress of
urban and mountain development all over Japan during those years. Although there
are no reliable data for damages, or even for casualties due to landslides in many coun-
tries of the world, Fig. 0.1 provides a clear evidence of the strong negative impact of
landslides to society. One clear difference of landslides, as compared to earthquakes
and volcanic eruptions, is that humans can prevent or mitigate many landslide phe-
nomena, while earthquakes and volcanic eruptions can not be prevented. The hazard
assessment of landslides is very effective for disaster reduction because of relatively
small affected area comparing to earthquakes and typhoons. Therefore, we can do
much for landslide disaster mitigation. In addition to human damages, landslides of-
ten destroy cultural and natural heritage, and the natural environment, which cannot

Fig. 0.1.
Comparison of the numbers
of victims in Japan from
19672004 due to landslide
disasters, earthquake disas-
ters including deaths by
slides, and volcanic disasters
including deaths due to volca-
nic gas (The data on victims
due to landslide disasters
since 1967 were published by
the Sabo Technical Center)
XIV Preface Aims of This Volume

be recovered. Landslide disaster reduction, protection of cultural and natural heritage

sites, and the invaluable environment are vital factors for human society. However,
studies of landslides have not been conducted in an integrated manner, although the
phenomena are targets of many scientific and engineering fields. Landsides are a tar-
get of application of many technical fields, but they are not the major interest of any
individual field. Therefore, no national landslide society has been established, except
the Japan Landslide Society (JLS), and the very recent Nepal Landslide Society. No
international society on landslides has been established because there are only one or
two national societies.
The Japan Landslide Society was founded in 1963, and the society organized the
first International Symposium on Landslides (ISL) in Kyoto in 1972, and the second
International Symposium on Landslides in Tokyo in 1977. The Japan Landslide Soci-
ety also initiated the International Conference and Field Workshop on Landslides (ICFL)
in Tokyo, 1985. The society started to publish the International Newsletter Landslide
News in 1987. Five thousand copies were printed in three colors and 2 000 copies were
distributed free of charge over the world. Landslide News continued to be published
until 2003 with supports from the United Nations Educational, Scientific and Cultural
Organization (UNESCO), the Food and Agriculture Organization of the United Na-
tions (FAO), the United Nations Secretariat for International Strategy for Disaster Re-
duction (UN/ISDR). This long-term activity gradually built up an infrastructure of
international landslide community. It led to the formation of the International Con-
sortium on Landslides (ICL) in 2002 and the new journal of ICL, Landslides, in 2004.
The publication of journal is the essential and necessary platform to create an inde-
pendent field of science dealing with landslides. This book Progress in Landslide
Science aims to present an overview of the current status of research by major land-
slide researchers worldwide. This volume can be a part of series of publications of ICL
that may contribute to the education system of landslide science and landslide disas-
ter management in the future.
The International Consortium on Landslides (ICL), United Nations Educational,
Scientific and Cultural Organization (UNESCO), World Meteorological Organization
(WMO), Food and Agriculture Organization of the United Nations (FAO), United Na-
tions International Strategy for Disaster Risk Reduction (UN/ISDR), United Nations
University (UNU), United Nations Environment Programme (UNEP), United Nations
Development Programme (UNDP), World Bank (IBRD), International Council for Sci-
ence (ICSU), World Federation of Engineering Organizations (WFEO), Kyoto Univer-
sity (KU), and the Japan Landslide Society (JLS) have jointly been organizing the First
World Landslide Forum for Strengthening Research and Learning on Earth System
Risk Analysis and Sustainable Disaster Management within the UN-ISDR as Regards
Landslides which will be held in November 2008 in Tokyo. This forum will be a
milestone in developing research and learning on landslide disaster reduction. Tan-
gible progress in Landslide Science is foreseen during this process. It is hoped that this
volume will be read by partners working for landslide risk mitigation and that further
works will be published as increasingly wider support is obtained.

Kyoji Sassa
Kyoto, January 2007
Landslides and Cultural Heritage:
the Common Thread from IGCP-425 to ICL/IPL

The protection of cultural heritage from natural hazards is an issue of worldwide con-
cern that regards advanced nations and developing countries alike. The history and
identity of every people is preserved in and portrayed by the unique monuments, art,
literature, archeology and artefacts that have been created during its existence. The
destruction of this wealth would represent an irreparable loss for the whole of human-
ity. The damage caused by natural disasters is increasing steadily as both the vulner-
ability of rapidly developing urban areas and the consequences of climate change tend
to amplify the effects caused by their occurrence.
Within this framework landslides represent a major threat both for the safety of
people and for the preservation of the built environment, including many important
sites of renowned international cultural or natural value. For instance, the Mediterra-
nean area is an excellent example of this situation as the immeasurable wealth of the
entire region is often exposed to landslide hazard due to its particular topographic,
geologic and climatic settings. Although the concept of preservation has already taken
hold in many of the European nations bordering on the Mediterranean, much still has
to be accomplished as the awareness of both the cultural and economic implications
of loss is only just starting to be appreciated. The situation is rather more serious in
developing countries, where even minimum levels of both the awareness of the unique
value represented by cultural heritage and the economic, scientific and technical means
for mitigating landslide hazard are often lacking.
For these reasons, the International Geological Correlation Programme (IGCP)
project 425 titled Landslide hazard assessment and mitigation for cultural heritage
sites and other locations of high societal value was approved during the 26th Session
of the IGCP Scientific Board. The main objectives of IGCP-425, proposed in 1998 by
Kyoji Sassa, were to assess and mitigate the effects of natural hazards on the cultural
heritage in countries from around the world and to propose reliable, cost-effective slope
stabilization techniques suitable for use on a global scale, including developing countries.
Before ending in 2002, IGCP-425 strengthened previous projects for the preserva-
tion of cultural heritage sites carried out under the Safeguarding Campaigns of
UNESCO, in which it played a key role in protecting 26 sites of global relevance. The
IGCP-425 was set up to provide a framework for assembling the results of national
projects carried out by the individual participants (over 50 national and regional in-
stitutions and universities were involved) concerning various aspects of landslide haz-
ard that are still useful today for drawing up prevention and mitigation plans of gen-
eral validity.
In particular, the specific aims of the IGCP-425 project were:

 to develop high precision and durable slope monitoring equipment for potential
landslides in urban and rural areas, especially in relation to outstanding cultural
heritage sites;
 to develop reliable, cost-effective techniques for assessing rapid landslide motion,
delimit hazard areas and reduce damage to life, cultural heritage, infrastructures
and property;
XVI Landslides and Cultural Heritage: the Common Thread from IGCP-425 to ICL/IPL

 to develop economical and effective slope stabilization works and disaster mitiga-
tion measures suitable for use on a global scale, including developing countries;
 to research reliable landslide hazard assessment and risk evaluation methods;
 to detect potential landslides and identify precursory phenomena.

To promote the project, UNESCO and the Disaster Prevention Research Institute of
Kyoto University, Japan (DPRI/KU) exchanged in 1999 a Memorandum of Understand-
ing concerning cooperation in research for landslide risk mitigation and protection of
the cultural and natural heritage as a key contribution to environmental protection
and sustainable development in the first quarter of the twenty-first century.
The Tokyo Declaration Geoscientists tame landslides was released in the 2001
UNESCO/IGCP Symposium on Landslide Risk Mitigation and Protection of Cultural
and Natural Heritage to propose an initiative that led to the creation of the Interna-
tional Consortium on Landslides (ICL) for the worldwide promotion of landslide re-
This initiative derived from two observations: (1) that the specific points listed pre-
viously correspond to the main themes of landslide research in general and (2) that
the specificity of the protection of cultural heritage from landslides must necessarily
take into account not only landslide research methods but also those relative to the
study of the vulnerability and worth of the cultural and natural artifact, which entails
the involvement of other disciplines and professions. These factors led to a practical
approach in which landslides are tackled comprehensively and cultural heritage is
considered as a related, highly important field.
In 2002, during the International Symposium Landslide Risk Mitigation and Pro-
tection of Cultural and Natural Heritage, co-organized by UNESCO and Kyoto Uni-
versity, international experts coming from different national, scientific, and govern-
mental institutes, academic institutions, regional and international organizations, in-
ternational non-governmental organizations and United Nations organizations unani-
mously agreed and declared to launch the ICL in the 2002 Kyoto Declaration. In the
months following its launch, a great effort by the IGCP-425 leaders led to the definition
of the objectives and the structure of the Consortium, and the results achieved during
the projects were adopted as the basis for the construction of a world research pro-
gram on landslides (International Programme on Landslides, IPL) as one of the main
initiatives of the recently created ICL.
The IPL is today a pillar of the ICL and both are rooted in the initial endeavor:
IGCP-425, the objectives of which are still the guiding principles. The IPL also gave
birth to the journal Landslides, indeed one of its most important outcomes, which
hosts many papers dedicated to the topic of landslides and cultural heritage.

Paolo Canuti
Firenze, January 2007

Part I
Progress in Landslide Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1 Landslide Science as a New Scientific Discipline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.1 Definition of Landslides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Landslide Science as a New Scientific Discipline
and Landslide Dynamics as its Core . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Foundation of the International Consortium on Landslides . . . . . . . . . . . . . . . . . . . . 6
1.4 Development of the International Landslide Community:
the 2005 Letter of Intent, the 2006 Tokyo Action Plan,
and the 2008 First World Landslide Forum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
The First World Landslide Forum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2 An Overview of Landslide Problems in the British Isles,

with Reference to Geology, Geography and Conservation . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Strong Rocks and Discontinuity-controlled Slope Instability . . . . . . . . . . . . . . . . . . 14
2.3 Weak Rocks and Strong Soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Erosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5 The Coastline of South Eastern Britain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.6 Bedding-Control of Slip Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.7 Slides in Strata with Low Angle Dips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.8 Grabens and Graben Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.9 Perched Slip Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.10 Slip Surfaces at or close to the Base of a Slope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.11 Identification of Slip Surface Position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.12 Three Dimensions and the Plan Shape of Landslides . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.13 Conservation and Conflict . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3 Considerations about the Mechanics of Slow Active Landslides in Clay . . . . . . . . . . . . . 27

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 General Features of Slow Active Landslides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3 Considerations about the Mechanics of Active Slides and Mudslides . . . . . . . . 28
3.4 Consideration about the Mechanics of Active Lateral Spreads . . . . . . . . . . . . . . . . . 39
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4 Dynamics of Rapid Landslides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2 Mechanisms Causing Strength Loss in Landslides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
XVIII Contents

4.3 Types of Extremely Rapid Landslides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5 Progress in Debris Flow Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.2 Processes of Initiation and Development
in the Erosion Type Debris Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.3 Criteria of Debris Flow Occurrence for Erosion Type . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.4 Processes in Transformation and Stoppage
of Landslide-induced Debris Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.5 Simplified Mathematical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.6 Models for Debris Flow Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.7 Single-phase Continuum Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.8 Two-phase Fluid Flow Model (Mixture Theory) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Part II
Landslide Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

6 Undrained Stress-controlled Dynamic-loading Ring-shear Test

to Simulate Initiation and Post-failure Motion of Landslides . . . . . . . . . . . . . . . . . . . . . . . . 81
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.2 Comparison with Triaxial Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.3 Structure and Control System of Apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.4 Testing Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.5 Undrained Shear Behavior on Sands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.6 Geotechnical Simulation of Earthquake-induced Landslides . . . . . . . . . . . . . . . . . . 91
6.7 Slide-triggered Debris Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.8 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

7 Shear Behavior and Shear Zone Structure of Granular Materials

in Naturally Drained Ring Shear Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7.2 Ring Shear Apparatus and Observation System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
7.3 Samples and Their Physical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
7.4 Testing Conditions and Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
7.5 Test Results on Mixed Sands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

8 Rockslides and Their Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

8.1 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
8.2 What is a Rockslide? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
8.3 Size Distribution from Grain Crushing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
8.4 Block Slides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
8.5 The Waikaremoana Blockslide Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
8.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Contents XIX

9 Residual Shear Strength of Tertiary Mudstone and Influencing Factors . . . . . . . . . . 135

9.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
9.2 Properties of Tertiary Mudstone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
9.3 Shear Strength Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
9.4 Factors Influencing the Residual Shear Strength of Mudstone . . . . . . . . . . . . . . 143
9.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

10 On Failure of Municipal Waste Landfill . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
10.2 Failure of Waste Leuwigajah Landfill in Bandung, Indonesia . . . . . . . . . . . . . . . . 147
10.3 Significance of Landfill Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
10.4 Proposals for Better Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
10.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

11 Experimental Study with Ring Shear Apparatus on the May 2004

LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan . . . . . . . . 151
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
11.2 General Conditions of the Jinnosuke-dani Landslide on Haku-san Mountain . 154
11.3 The May 2004 LandslideDebris Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
11.4 Ring Shear Tests on the Initiation and Traveling Mechanisms
of the LandslideDebris Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
11.5 Ring-shear Tests on Soil Samples Taken from
the Landslide Travel Path in the Bettou-dani . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
11.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

12 On the Pore-pressure Generation and Movement

of Rainfall-induced Landslides in Laboratory Flume Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
12.2 Properties of the Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
12.3 Flume Test Apparatus and Test Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
12.4 Observed Phenomena in Flume Tests and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 171
12.5 Double-cylinder Apparatus and Test Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
12.6 Pore-pressure-maintaining Mechanism during Movement . . . . . . . . . . . . . . . . . . 178
12.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180

13 Ring Shear Tests on Clays of Fracture Zone Landslides

and Clay Mineralogical Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
13.2 Features of Fracture Zone Landslides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
13.3 Experimental Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
13.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
13.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192

14 Landslides Induced by a Combined Effect of Earthquake and Rainfall . . . . . . . . . . . . 193

14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
14.2 Combined Effect of Rainfall and Earthquake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
XX Contents

14.3 The 2006 Leyte Landslide Triggered by a Small Near-by Earthquake (M2.6)
Five Days after a Heavy Rainfall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
14.4 The 2004 Higashi-Takezawa Triggered by the Magnitude (M6.8)
Earthquake Three Days after the Typhoon No. 23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
14.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206

15 Landslide Experiments on Artificial and Natural Slopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209

15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
15.2 Landslide Fluidization Process by Flume Experiment . . . . . . . . . . . . . . . . . . . . . . . . 210
15.3 A Fluidized Landslide on Natural Slope Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

Part III
Landslide Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227

16 Enlargement of a Failed Area along a Sliding Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229

16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
16.2 Field Monitoring of Deformation Area Enlargement
in a Landslide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
16.3 Study of the Formation of Sliding Surface in Laboratory Tests . . . . . . . . . . . . . . 232
16.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236

17 Airborne LIDAR Data Measurement and Landform Classification Mapping

in Tomari-no-tai Landslide Area, Shirakami Mountains, Japan . . . . . . . . . . . . . . . . . . . . 237
17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
17.2 Study Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
17.3 Airborne LIDAR Data Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
17.4 Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
17.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
17.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

18 Integration of Remote Sensing Techniques in Different Stages

of Landslide Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
18.2 Contributions to Landslide Inventories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
18.3 Improvements to Landslide Hazard Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
18.4 Basin Scale Risk Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
18.5 Landslide Modeling Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
18.6 Landslide Monitoring by Remote Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
18.7 Innovative Early Warning Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
18.8 Support for Emergency Management and Scenario Analysis . . . . . . . . . . . . . . . . 258
18.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259

19 Rock Deformation Monitoring at Cultural Heritage Sites in Slovakia . . . . . . . . . . . . . 261

19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
19.2 Works and Techniques Employed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
19.3 Spis Castle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
19.4 Strecno Castle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
Contents XXI

19.5 Skalka Monastery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269

19.6 Lietava Castle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
19.7 Lednica Castle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
19.8 C`achticky hrad Castle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
19.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273

Part IV
Landslide Risk Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275

20 Extracting Necessary Parameters from Real Landslide Mass

for Mitigating Landslide Disaster . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
20.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
20.2 Estimation of Velocity of Fluidized Soil Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
20.3 Extracting Parameters for Estimating Travel Distances
for Coherent Mass Movement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
20.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284

21 Landslide Dams Formed by the 2004 Mid-Niigata Prefecture Earthquake

in Japan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
21.2 Characteristics of the 2004 Mid-Niigata Prefecture Earthquake . . . . . . . . . . . . 285
21.3 Distribution of Landslides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
21.4 River Blockage by Landslide Dam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
21.5 Emergency Operations against Collapse by Landslide Dam . . . . . . . . . . . . . . . . . 290
21.6 Monitoring and Observation System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
21.7 Successive Landslides Caused by Snowmelt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
21.8 Remarks for Mitigation of Future Disasters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293

22 Shear Behavior of Clay in Slope for Pore Water Pressure Increase . . . . . . . . . . . . . . . . . 295
22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
22.2 Pore Water Pressure Loading Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
22.3 Consideration on Landslide Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
22.4 Stress Controlled Ring Shear Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
22.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303

23 Static and Dynamic Analyses of Slopes by the FEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305

23.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
23.2 Finite Element Analysis of Slopes and Slope Stabilization . . . . . . . . . . . . . . . . . . . . 305
23.3 Dynamic Elasto-plastic Finite Element Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
23.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311

24 Debris Flows in the Vicinity of the Machu Picchu Village, Peru . . . . . . . . . . . . . . . . . . . . . 313
24.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
24.2 Preliminary Field Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
24.3 Triggering Factors of the Debris Flows in the Machu Picchu Village
and Surrounding Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
XXII Contents

25 Engineering Geology and Cultural Heritage:

the Conservation of Remaining Bamiyan Buddhas (Central Afghanistan) . . . . . . . . 319
25.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
25.2 Meteo-climatic Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
25.3 Geological, Mineralogical and Petro-geophysical Setting . . . . . . . . . . . . . . . . . . . . 321
25.4 Scanning Electron Microscopy (SEM) of Siltstone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
25.5 Physical and Mechanical Properties of Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
25.6 Geomorphological Setting and Most Unstable Areas . . . . . . . . . . . . . . . . . . . . . . . . . 326
25.7 Seismological Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
25.8 Geomechanic Characters of Discontinuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
25.9 Structural Analysis of Discontinuities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
25.10 Kinematic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
25.11 Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
25.12 Previous Restoration Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
25.13 Long-term Conservation Strategy for Repair, Enhancement,
Research, and Risk-preparedness for the Preservation of the Site . . . . . . . . . . 333
25.14 Identification of Most Unstable Areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
25.15 Emergency Measures in the Upper Eastern Part
of the Eastern Giant Buddha Niche . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336
25.16 Completion of Emergency Intervention
in the Eastern Giant Buddha Niche . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
25.17 First Interventions in the Western Giant Buddha Niche . . . . . . . . . . . . . . . . . . . . . . 341
25.18 The Back Side of Both Buddha Niches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
25.19 Manual Crack Gauge Monitoring System: Current Results . . . . . . . . . . . . . . . . . . . 344
25.20 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346

26 Debris Flow Hazard Defense Magnitude Assessment with Numerical Simulation . . . 347
26.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
26.2 Literature Review on Flood Hazard Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
26.3 Assessment of Debris Flow Hazard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
26.4 Land Utilization within the Influenced Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
26.5 The Expected Disaster Loss in the Influenced Area . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
26.6 Case Study Assessment for Taipei 021 Debris Flow Potential Stream . . . . . 355
26.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360

Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363

A.1 The Tokyo Action Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363

A1.1 Preamble . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
A1.2 Action Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
A1.3 After the 2006 Tokyo Action Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366

A.2 MoUs between ICL and Global Stakeholders to Promote

the 2006 Tokyo Action Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
A2.1 MoU between ICL and UN/ISDR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
A2.2 MoU between ICL and UNU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
A2.3 MoU between ICL and WMO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
A2.4 MoU between ICL and WFEO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
A2.5 MoU between ICL and UNESCO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
A2.6 MoU between ICL and ICSU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
List of Contributors

Araiba, Kiminori (Chapter 16)

National Research Institute of Fire and Disaster, Jindaiji Higashimachi 4-35-3, Chofu, Tokyo 182-8508,
Corresponding author of Chapter 16:
Tel: +81-422-44-8331, Fax: +81-422-42-7719, E-mail: araiba@fri.go.jp

Bhandary, Netra P. (Chapter 13)

Graduate School of Science and Engineering, Ehime University, Bunkyo-3, Matsuyama 790-8577,
Corresponding author of Chapter 13:
Tel: +81(0)89-927-8566, Fax: +81(0)89-927-8566, E-mail: netra@dpc.ehime-u.ac.jp

Bromhead, Edward N. (Chapter 2)

School of Engineering, Kingston University, Penrhyn Road, Kingston upon Thames, KT1 2EE, UK
Corresponding author of Chapter 2:
Tel: +44-20-8547-2000 ext 62225, E-mail: e.bromhead@kingston.ac.uk

Cai, Fei (Chapter 23)

Graduate School of Eng., Gunma University, Tenjin-cho 1-5-1, Kiryu 376-8515, Japan

Canuti, Paolo (Chapter 18)

Department of Earth Sciences, University of Firenze, Via Giorgio La Pira 4, 50121 Firenze, Italy

Casagli, Nicola (Chapter 18)

Department of Earth Sciences, University of Firenze, Via Giorgio La Pira 4, 50121 Firenze, Italy
Corresponding author of Chapter 18:
Tel: +39 055 2757523, Fax: +39 055 2756296, E-mail: nicola.casagli@unifi.it

Catani, Filippo (Chapter 18)

Earth Sciences Department, University of Firenze, Via Giorgio La Pira 4, 50121 Firenze, Italy

Davies, Tim (Chapter 8)

Department of Geological Sciences, University of Canterbury, Christchurch, New Zealand

Falorni, Giacomo (Chapter 18)

Earth Sciences Department, University of Firenze, Via Giorgio La Pira 4, 50121 Firenze, Italy

Farina, Paolo (Chapter 18)

Department of Earth Sciences, University of Firenze, Via Giorgio La Pira 4, 50121 Firenze, Italy
XXIV List of Contributors

Fukuoka, Hiroshi (Chapters 6, 7, 14)

Research Centre on Landslides, Disaster Prevention Research Institute, Kyoto University, Gokasho,
Uji, Kyoto 611-0011, Japan
Corresponding author of Chapter 7:
Tel: +81-774-38-4111, Fax: +81-774-38-4300, E-mail: fukuoka@scl.kyoto-u.ac.jp

Greif, Vladimir (Chapter 19)

Department of Engineering Geology, Comenius University Bratislava, Slovakia
Corresponding author of Chapter 19:
Tel: +421-2-6029-6627, Fax: +421-2-6029-6702, E-mail: greif@nic.fns.uniba.sk

Hencelova, Lucia (Chapter 19)

Department of Engineering Geology, Comenius University Bratislava, Slovakia

Holzer, Rudolf (Chapter 19)

Department of Engineering Geology, Comenius University Bratislava, Slovakia

Hsu, Yu-Charn (Chapter 26)

PhD Candidate, Department of Civil Engineering, National Taiwan University

Hungr, Oldrich (Chapter 4)

Department of Earth and Ocean Sciences, University of British Columbia, 6339 Stores Rd., Vancouver,
B.C. V6T 1Z4, Canada
Corresponding author of Chapter 4:
Tel: +1-604-822-8471, Fax: +1-604-822-6088, , E-mail: ohungr@eos.ubc.ca

Ibsen, Maia L. (Chapter 2)

School of Engineering, Kingston University, Penrhyn Road, Kingston upon Thames, KT1 2EE, UK

Iwahashi, Junko (Chapter 17)

Geography and Crustal Dynamics Research Center, Geographical Survey Institute, Ministry of Land,
Infrastructure and Transport, 1 Kitasato, Tsukuba, Ibaraki 305-0811, Japan

Jezny, Michal (Chapter 19)

Department of Engineering Geology, Faculty of Natural Sciences, Comenius University Bratislava,
84215 Bratislava, Mlynska dolina, Slovak Republic

Johansson, Jrgen (Chapter 20)

Institute of Industrial Science, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan

Klime, Jan(Chapter 24)

Institute of Rock Structure and Mechanics, Academy of Sciences, V Holeovikch 41, 182 09 Prague 8,
Czech Republic

Koarai, Mamoru (Chapter 17)

Geography and Crustal Dynamics Research Center, Geographical Survey Institute, Ministry of Land,
Infrastructure and Transport, 1 Kitasato, Tsukuba, Ibaraki 305-0811, Japan

Konagai, Kazuo (Chapter 20)

Institute of Industrial Science, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan
Corresponding author of Chapter 20:
Tel: +81-3-5452-6142, Fax: +81-3-5452-6144, E-mail: konagai@iis.u-tokyo.ac.jp
List of Contributors XXV

Li, Hsin-Chi (Chapter 26)

Researcher, socio-economic system division, National Science & Technology Center for Disaster

Liu, Ko-Fei (Chapter 26)

Department of Civil Engineering, National Taiwan University, Slope Land Disaster Reduction
Division, National Science & Technology Center for Disaster Reduction
Corresponding author of Chapter 26:
Tel: +886-2-2365-5405, Fax: +886-2-2363-1558, E-mail: kfliu@ntu.edu.tw

Margottini, Claudio (Chapter 25)

ENEA CR Casaccia, Via Anguillarese 301, 00060 S. Maria di Galeria, Rome, Italy
Corresponding author of Chapter 25:
Tel: +39-06-3048-4688, Fax: +39-06-3048-4029, E-mail: margottini@casaccia.enea.it

Marui, Hideaki (Chapter 21)

Research Center for Natural Hazards and Disaster Recovery, Niigata University,
8050 Ikarashi-Ninocho, Niigata, 950-2181, Japan
Corresponding author of Chapter 21:
Tel: +81-25-262-7055, Fax: +81-25-262-7050, E-mail: maruihi@cc.niigata-u.ac.jp

McSaveney, Mauri (Chapter 8)

GNS Science, Lower Hutt, New Zealand
Corresponding author of Chapter 8:
Tel: +64-4-570-4734, Fax: +64-4-570-4600, E-mail: M.McSaveney@gns.cri.nz

Numada, Muneyoshi (Chapter 20)

Institute of Industrial Science, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505,

Ochiai, Hirotaka (Chapter 15)

Forestry and Forest Products Research Institute, 1 Matsunosato, 305-8687 Tsukuba, Ibaraki,
Corresponding author of Chapter 15:
Tel: +81-29-829-8237, Fax: +81-29-874-3720, E-mail: ochi@ffpri.affrc.go.jp

Ohtsuka, Satoru (Chapter 22)

Department of Civil & Environmental Engineering, Nagaoka University of Technology,
1603-1 kamitomioka-cho, Nagaoka, Niigata 940-2199, Japan
Corresponding author of Chapter 22:
Tel: +81-258-47-9633, Fax: +81-258-47-9600, E-mail: ohtsuka@nagaokaut.ac.jp

Okada, Yasuhiko (Chapter 15)

Forestry and Forest Products Research Institute, 1 Matsunosato, 305-8687 Tsukuba, Ibaraki,

Picarelli, Luciano (Chapter 3)

Dipartimento di Ingegneria Civile, Seconda Universit di Napoli, Aversa, Italy
Corresponding author of Chapter 3:
Tel: +39-081-5010213, Fax: +39-081-5037370, E-mail: luciano.picarelli@unina2.it

Sammori, Toshiaki (Chapter 15)

Forestry and Forest Products Research Institute, 1 Matsunosato, 305-8687 Tsukuba, Ibaraki,
XXVI List of Contributors

Sassa, Kyoji (Chapters 1, 6, 7, 11, 12, 14)

Research Centre on Landslides, Disaster Prevention Research Institute, Kyoto University, Uji,
Kyoto 611-0011, Japan
Corresponding author of Chapters 1, 6, 14:
Tel: +81-774-38-4110, Fax: +81-774-32-5597, E-mail: sassa@scl.kyoto-u.ac.jp

Sato, Hiroshi P. (Chapter 17)

Geography and Crustal Dynamics Research Center, Geographical Survey Institute, Ministry of Land,
Infrastructure and Transport, 1 Kitasato, Tsukuba, Ibaraki 305-0811, Japan
Corresponding author of Chapter 17:
Tel: +81-29-864-5946, Fax: +81-29-864-2655, E-mail: hsato@gsi.go.jp

Sekiguchi, Tatsuo (Chapter 17)

Geography and Crustal Dynamics Research Center, Geographical Survey Institute, Ministry of Land,
Infrastructure and Transport, 1 Kitasato, Tsukuba, Ibaraki 305-0811, Japan

Suemine, Akira (Chapter 16)

Research Centre on Landslides, Disaster Prevention Research Institute, Kyoto University, Gokasho,
Uji, Kyoto 611-0011, Japan

Takahashi, Tamotsu (Chapter 5)

Prof. Emeritus of Kyoto University, 3-6-26, Kita-Oji, Otsu, Shiga 520-0843, Japan
Corresponding author of Chapter 5:
E-mail: ta-taka@iris.eonet.ne.jp

Tiwari, Binod (Chapter 9)

Department of Civil and Environmental Engineering, College of Engineering and Computer Sciences,
California State University, Fullerton, 800 N. State College Blvd. E-419, Fullerton, CA 92834, U.S.A.
Corresponding author of Chapter 9:
Tel: +1-714-278-3968, Fax: +1-714-278-3916, E-mail: btiwari@fullerton.edu

Towhata, Ikuo (Chapter 10)

Department of Civil Engineering, University of Tokyo, 7-3-1, Hongo, Bunkyo-Ku, Tokyo 113-8656,
Corresponding author of Chapter 10:
Tel: +81-3-5841-6121, Fax: +81-3-5841-8504, E-mail: towhata@geot.t.u-tokyo.ac.jp

Ugai, Keizo (Chapter 23)

Graduate School of Eng., Gunma University, Tenjin-cho 1-5-1, Kiryu 376-8515, Japan

Vilmek,Vt (Chapter 24)

Department of Physical Geography and Geoecology, Faculty of Science, Charles University,
Albertov 6, 128 43 Prague 2, Czech Republic
Corresponding author of Chapter 24:
Tel: +420-22-0561-361, Fax: +420-22-1951-367, E-mail: vilimek@natur.cuni.cz

Vlko, Jn (Chapter 19, 24)

Department of Engineering Geology, Faculty of Natural Sciences, Comenius University Bratislava,
Mlynska dolina G, 842 15 Bratislava, Slovakia

Wakai, Akihiko (Chapter 23)

Graduate School of Eng., Gunma University, Tenjin-cho 1-5-1, Kiryu 376-8515, Japan
Corresponding author of Chapter 23:
Tel: +81-277-30-1624, Fax: +81-277-30-1624, E-mail: wakai@ce.gunma-u.ac.jp
List of Contributors XXVII

Wang, Fawu (Chapters 6, 11, 14)

Research Centre on Landslides, Disaster Prevention Research Institute, Kyoto University, Gokasho,
Uji, Kyoto 611-0011, Japan
Corresponding author of Chapter 11:
Tel: +81-774-38-4114, Fax: +81-774-38-4300, E-mail: wangfw@landslide.dpri.kyoto-u.ac.jp

Wang, Gonghui (Chapters 6, 7, 12, 14)

Research Centre on Landslides, Disaster Prevention Research Institute, Kyoto University, Japan, Uji,
Kyoto, 611-0011, Japan
Corresponding author of Chapter 12:
Tel: +81-774-38-4114, Fax: +81-774-38-4300, E-mail: wanggh@landslide.dpri.kyoto-u.ac.jp

Yagi, Hiroshi (Chapter 17)

Yamagata University, Japan

Yatabe, Ryuichi (Chapter 13)

Graduate School of Science and Engineering, Ehime University, Bunkyo-3, Matsuyama 790-8577,

Yoshimatsu, Hiroyuki (Chapter 21)

Sabo Technical Center, 4-8-1, Kudan-minami, Chiyoda-ku Tokyo, 102-0074, Japan
Part I Progress in
Landslide Science

Chapter 1 Landslide Science as a New Scientific Discipline

Chapter 2 An Overview of Landslide Problems in the British Isles,

with Reference to Geology, Geography and Conservation

Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay

Chapter 4 Dynamics of Rapid Landslides

Chapter 5 Progress in Debris Flow Modeling

Chapter 1

Landslide Science as a New Scientific Discipline

Kyoji Sassa

Abstract. Landslides cause great disasters and their impact to soci- Some landslides are triggered by human activities, such
ety is very great. Thus, they are studied in many scientific and engi- as road and railway construction, mining, and develop-
neering fields. However, studies on landslides from various fields
ment in urban and mountain areas. The landslide phe-
have not been conducted in an integrated manner. There was nei-
ther international society, nor international journal, and the mean- nomena have been studied in may countries and many
ing of landslides was not defined internationally and interdiscipli- areas. Landslide studies have been conducted in many
nary. During the United Nations International Decade for Natural fields of science and engineering. Landslide disasters have
Disaster Reduction (IDNDR) in 19902000, landslide researchers been dealt with many governments, ministries, and they
worldwide agreed the definition of landslides as the movement of
are related to many other disasters. As shown in Fig. 1.1,
a mass of rock, debris or earth down a slope. This is a basis for the
development of the study of landslides as a scientific field. This pa- landslides are phenomena that involve many disciplines,
per describes the progress in landslide science as an integrated dis- ministries, and individuals. However, landslide phenom-
cipline together with the development of international landslide ena have not been studied in an integrated way. Even the
community and a global cooperation platform as its infrastructure. definition of landslides, what is a landslide, has been
diverse in countries and disciplines worldwide.
Keywords. Classification of landslides, landslide dynamics, landslide
science, ISDR, IDNDR, ICL
David Varnes presented Classification of Landslides
by the type of material (bedrock and soils) and type of
movement (falls, slides, flows) in Landslides and Engi-
neering Practice (Eckel, ed.) in 1958 (Varnes 1958). This
1.1 Definition of Landslides paper presented the concept of landslides in wider mean-
ing, including debris flows, rock falls, debris avalanches,
Landslides are studied in many countries because they creep, etc. However, many opinions and criticism occurred
occur almost worldwide, from high mountain areas to whether land-SLIDE can include the phenomena of FLOW,
coastal areas and even in marine geologic units, from very FALL and other movements. Therefore, Varnes (1978)
wet or heavy rainfall areas to very dry areas, and from presented Types of Slope Movements in Landslides-
seismic or volcanic areas to tectonically non-active areas. Analysis and Control using the type of material (bed-

Fig. 1.1. Characteristics of landslide disasters (from 2006 Tokyo Action Plan; Sassa 2006)
4 Kyoji Sassa

rock, debris, earth) and the type of movement (falls, topples, Landslides were classified by the Working Party by the
slides, lateral spreads, flows; Fig. 1.2). The International type of material (rock, debris: predominantly coarse soils,
Geotechnical Societies UNESCO Working Party on World and earth: predominantly fine soils) and the type of move-
Landslide Inventory was established in conjunction with ment (fall, topple, slide, spread, flow). Basically the idea
the United Nations International Decade for Natural Disas- was to return to the Varnes Classification of Landslides
ter Reduction (IDNDR). This Working Party was formed (Varnes 1958). It was introduced by Cruden and Varnes
by the Commission on Landslides and other Mass Move- (1996) in Landslides Investigation and Mitigation. The
ments of the International Association of Engineering Ge- definition of landslides as the the movement of a mass
ology and the Environment (IAEG), the Technical Commit- of rock, debris or earth down a slope was widely accepted.
tee on Landslides of the International Society for Soil Me- It is the very important basis for the development of Land-
chanics and Geotechnical Engineering (ISSMGE) and the slide Science.
International Society for Rock Mechanics (ISRM). This The International Consortium on Landslides (ICL)
working party defined various terms of landslide features, which was established in 2002 as the first and unique in-
landslide velocities, landslide dimensions, and state of ac- ternational organization dedicated to landslide research,
tivities (Cruden and Varnes 1996). The most important of created a new award called the Varnes Medal, which
these factors is the definition of landslides by the Working recognizes professional excellence in landslide research
Parties as the movement of a mass of rock, debris or earth and education, the basis for David Varnes definition of
down a slope. Namely various types of gravitational mass the area of landslide study. The first Varnes Medal was
movements were integrated in the category of Landslides. bestowed to Robert Schuster of the U.S. Geological Sur-
Even some movements are not Slide. Cruden (1991) ex- vey (Canuti 2004).
plained that an English word combining two words can ex-
press something different from original two words. Thus,
landslide is not always necessary to be Slide of Land. 1.2 Landslide Science as a New Scientific
Discipline and Landslide Dynamics as its Core

The International Strategy for Disaster Risk Reduction

of the United Nations (UN-ISDR) was initiated in 2000
following the International Decade for Natural Disaster
Reduction (IDNDR) for 19902000. Landslide disaster risk
reduction is one of major task of this strategy. Landslide

Fig. 1.2. Types of landslides (modified from Cruden and Varnes 1996) Fig. 1.3. Landslide Science as a new integrated discipline
Chapter 1 Landslide Science as a New Scientific Discipline 5

study by integrating knowledge obtained in many fields earthquakes (seismic wave loading up to 5 Hz in both
of science and engineering related to landslides is impera- normal stress and shear stress). Therefore, the appara-
tive to effectively mitigate landslide risk. Figure 1.3 pre- tuses can study pore-pressure generation and mobilized
sents an illustration of landslide study as a possible inte- shear resistance during sliding-surface formation and
grated discipline. However, It necessitates at least two fac- post-failure motion in various types of landslides; such
tors to be an independent discipline, namely its own core as rain-induced landslides, earthquake-induced land-
study in research and an international journal to bring slides, transformation of debris slides to debris flows, and
together common scientific knowledge. enlargement of landslide masses during the process of
downslope movement. Those applications were intro-
duced in the initial issue of the new international journal
1.2.1 Core Study of Landslide Science Landslides (Sassa et al. 2004). This study is one of core
studies in landslide science as a new discipline on land-
The core study in landslide research may develop as an slides. Within this initial issue of Landslides, a new moni-
integrated and independent discipline, namely landslide toring technology for landslides, the ground-based Syn-
science. One of the initial core studies in landslide sci- thetic Aperture Radar system (GB-SAR), was also reported
ence can be landslide dynamics because landslides have on. SAR was originally developed as a satellite monitor-
been defined as the Movement of a Mass of rock, debris ing method; however, this new system is installed on a
or earth down a slope. Dynamics are studied in the field short rail on the ground in front of landslides. The radar
of sciences dealing fluids: air, water, and other liquids. antenna is repeatedly moved on the short rail. Therefore,
Geology, geomorphology, and geotechnology dealing with monitoring for a short time span is possible (Antonello
solids on earth do not include dynamics, at least as their et al. 2004). Application of GIS and geophysical technol-
central interests. The stability analysis for the failure of ogy for landslide mapping and investigation, and appli-
slopes is a major task for geotechnology, but post-failure cation of stability technology to protect the Bamiyan
motion is not a concern because major interests are to Budda niches in Afghanistan were also reported in this
design construction works, such as roads, dams, and oth- initial issue.
ers, in a state that does not allow failure. The tools to study
dynamics of soil masses have not been fully developed.
Sassa (1992) developed an apparatus to simulate the ini- 1.2.2 A New International Journal, Landslides
tiation and post-failure motion of earthquake induced
landslides. This apparatus was not well functioned to The publication of an international journal is imperative
monitor pore-pressure generation in the shear zone and for an independent discipline. Such publication requires
to keep a shear box in the undrained condition during a stable financial background and a strong involvement
tests. The apparatus was later improved in its capability of the international scientific community to contribute
to maintain an undrained state in the soil sample and to quality articles. It is not easy to create a new international
monitor pore-pressure generation near the sliding sur- journal. The number of individuals, organizations, and
face. It was applied to the Nikawa landslide triggered by entities that are involved in landslide research are many,
the 1995 Kobe earthquake, which killed 34 residents by as is shown in Fig. 1.1. However, the total number of pro-
its rapid and long travel movement (Sassa 1996). By ob- fessionals studying landslides as their main interest is
taining a special budget to investigate the 1995 Kobe earth- small. The target of readers of new landslide journal must
quake disaster and to prevent further similar disasters, a be the wide variety of groups and disciplines shown in
new advanced undrained dynamic loading ring shear Fig. 1.1. Therefore, the articles of the journal must be en-
apparatus (DPRI-5 and 6) was developed (Sassa 2000). joyed by those people who have not professionally stud-
The apparatus is called as a ring-shear simulator of earth- ied and those who have no time to read in detail, but have
quake-induced landslides, with capability to reproduce time to glance at the articles. Thus, articles in the new
5 Hz real seismic wave loading, to maintain undrained journal must be something understood by a glance, and
condition in the soil sample during a maximum 224 cm s1 enjoyed by looking over pages of the journal. Ideally, the
velocity at the center of ring shear sample box (250 mm journal should be printed in full color. Mono-color land-
inside diameter, 350 mm outside diameter for DPRI-6). slide photos can neither be fully understood nor attrac-
The latest version is the ring shear apparatus (DPRI-7) tive. Figures must be drawn in color to present their mean-
which has a transparent shear box enabling the direct ing immediately in an understandable way. However, pub-
observation of shearing and crushing of grains at high lication of a full color journal is very expensive. As far as
speed under a high normal stress at a maximum speed the author knows, no full-color scientific journal has been
of 300 cm s1 (Sassa et al. 2004). These apparatuses, published without advertising, or without additional
DPRI-5,6,7, can reproduce the stress in the slope during charges for the color figures. Sassa proposed publication
ground-water rise (pore-pressure increase) or during of the journal Landslides as the initial and the life-time
6 Kyoji Sassa

project of the International Programme on Landslides Palace. An extensive landslide prevention work was con-
(IPL) at the first session of Board of Representatives of ducted to stabilize the slope. This work was funded at a
ICL held at UNESCO Headquarters on November 1921, level of three million U.S. dollars by municipal, regional,
2002. Publication of Landslides was decided to begin as and national governments of China. Probably this is the
the IPL C100 project. Sassa surveyed and negotiated with first case of the initiation of extensive landslide remedial
major international publication companies. Agreement measures at a cultural heritage site for mitigation of po-
was reached with Springer-Verlag at Heidelberg, Germany. tential landslides at the precursor stage in the world. This
Dr. Wolfgang Engel, the Executive Editor of Geosciences investigation of landslides at the precursor stage was
of Springer-Verlag, was the main partner. Landslides evaluated as a contribution of geoscientists to protection
began publication in April 2004 as a quarterly journal. of Cultural Heritage.
The journal was accepted at Thompson ISI for coverage Based on the invitation by Edward Derbyshire, Chair-
in Science Citation Index Expanded from 2005. It is ex- man of the Scientific Board of International Geological
tremely fast to be accepted in this index after only one Correlation Programme (IGCP), which is a joint program
year of publication. The journal is distributed through funded by UNESCO and the International Union of Geo-
the web at most of universities, institutes, and other or- logical Sciences (IUGS), and encouragement by Hideo
ganizations that have contracts to purchase web journals Noguchi (Programme specialist of the Division of Cultural
with Springer-Verlag. The printed version of the journal Heritage of UNESCO), Kyoji Sassa proposed an IGCP project
is distributed to ICL member organizations and others in 1998. Then, the UNESCOIUGS joint project, Interna-
through Springer-Verlag sales worldwide. It is said that tional Geological Correlation Programme (IGCP) No. 425,
the journal is fulfilling the initial purpose to attract groups, Landslide hazard assessment and mitigation for cultural
organizations, and individuals in many fields in both con- heritage sites and other locations of high societal value
tribution and reading. The number of pages submitted began. Thirty-one subprojects were proposed to join this
and the quality of papers are constantly increasing. IGCP-425 project worldwide. This project obtained about
4000 US dollars from UNESCO and IUGS. The budget was
shared to 31 subproject leaders as a part of their travel fees
1.3 Foundation of the International Consortium to attend the IGCP-425 meeting. This small amount of bud-
on Landslides get was very effective in promoting subproject leaders and
in raising fund in their countries. Because IGCP projects
Landslide science as an integrated discipline is developing have to be terminated within 5 years, the IGCP-425 group
through the journal publication and progress of core stud- wished to establish its own international program on land-
ies. As its background for the development, the International slides by creating an international organization on land-
Consortium on Landslide (ICL) and the recent development slides similar to the IUGS in IGCP program.
of the international landslide community under the initia- As a first step, a cooperative agreement was proposed
tive of ICL are introduced in this and the next section. between UNESCO and the institute of the IGCP-425 leader,
As a part of the Japanese contribution to the IDNDR the Disaster Prevention Research Institute, Kyoto Univer-
(International Decade for Natural Disaster Reduction) in sity. This proposal was made at the International Confer-
the last decade of the 20th century, the Ministry of Educa- ence Cultural Heritage at Risk, which was organized by
tion, Culture, Sports, Science and Technology of the Gov- UNESCO and the IGCP-425 group at UNESCO Headquar-
ernment of Japan (MEXT) conducted international joint ters, Paris on 2024 September 1999. Then, the Memo-
research projects. The projects included the Japan-China randum of Understanding (MoU) between the United
Joint Project Assessment of Landslide Hazards in Lishan Nations Educational, Scientific and Cultural Organization
(Yang-Que-Fe Palace), Xian, China, which was proposed (UNESCO) and the Disaster Prevention Research Insti-
by Kyoji Sassa, Disaster Prevention Research Institute, tute (DPRI), Kyoto University concerning Cooperation
Kyoto University. The Palace is an important Cultural in Research for Landslide Risk Mitigation and Protection
Heritage site, attracting more than three million visitors of the Cultural and Natural Heritage as a Key Contribu-
per year. The report of the joint research clearly demon- tion to Environmental Protection and Sustainable Devel-
strated evidence of the risk of large-scale rock slide, based opment in the First Quarter of the Twenty-First Century
on detailed monitoring and observation of two investi- was signed by Koichiro Matsuura, Director General of
gation tunnels. The Secretary-General of the Communist UNESCO, on 26 November and by Shuichi Ikebuchi, Di-
Party of the Shaanxi Provincial Government, the former rector of the Disaster Prevention Research Institute, Kyoto
director of the Chinese Seismological Bureau, and the University, on 3 December 1999. Based on this MoU,
honorary chairperson of the International Symposium on UNESCO and Kyoto University jointly organized the in-
Landslide Hazard Assessment, Xian, China organized by ternational symposium Landslide Risk Mitigation and
this group understood the landslide risk at the Lishan Protection of Cultural and Natural Heritage in Kyoto.
Chapter 1 Landslide Science as a New Scientific Discipline 7

Fig. 1.4. Group photo commemorating the establishment of the International Consortium on Landslides on 23 January 2005 at the Kyoto
Campus Plaza

Eight representatives of UNESCO, the World Meteorologi- 2. integrate geosciences and technology within the ap-
cal Organization (WMO), and the United Nations Secre- propriate cultural and social contexts in order to evalu-
tariat for the International Strategy for Disaster Reduc- ate landslide risk in urban, rural and developing areas
tion (UN/ISDR) participated in the symposium. The par- including cultural and natural heritage sites, as well as
ticipants decided to establish the ICL. The Statutes of the contribute to the protection of the natural environment
ICL were adopted and the first President (Kyoji Sassa) and and sites of high societal value;
interim steering committee members were nominated. By 3. combine and coordinate international expertise in
releasing the 2002 Kyoto Appeal Establishment of a New landslide risk assessment and mitigation studies,
International Consortium on Landslides, the International thereby resulting in an effective international organi-
Consortium on Landslides was inaugurated on 21 January zation, which will act as a partner in various interna-
2002. Figure 1.4 is a group photo of the participants. tional and national projects; and
The International Consortium on Landslides (ICL), 4. promote a global, multidisciplinary program on land-
created during the Kyoto Symposium in January 2002, is slides.
an international non-governmental and non-profit sci-
entific organization, which has been supported from the The central activity in the ICL is the International Pro-
beginning by the United Nations Educational, Scientific gramme on Landslides (IPL). The necessity of establish-
and Cultural Organization (UNESCO), the World Meteo- ment of a new Research Centre on Landslides to support
rological Organization (WMO), the Food and Agriculture IPL was proposed in the inaugural meeting of the ICL
Organization of the United Nations (FAO), the United on 21 January 2002. The new Research Centre on Land-
Nations International Strategy for Disaster Risk Reduc- slides (RCL) was established on 1 April 2003 in the Di-
tion (UN/ISDR), and intergovernmental programmes saster Prevention Research Institute, Kyoto University
such as the International Hydrological Programme of (DPRI/KU). The secretariat of the IPL is now hosted at
UNESCO, the International Union of Geological Sciences the headquarters building of UNESCOs UNITWIN (uni-
(IUGS), the Ministry of Education, Culture, Sports, Sci- versity twinning and networking) Cooperation Pro-
ence and Technology (MEXT) of the Government of Ja- gramme Landslide Risk Reduction for Society and the
pan, the U.S. Geological Survey, and other governmental Environment, which was constructed by Kyoto Univer-
bodies. The ICL was registered as a legal body under the sity and the International Consortium on Landslides on
Japanese law for non-profit organizations in August 2002 the Uji campus of Kyoto University in 2004. The rela-
by the Government of Kyoto Prefecture, Japan. tionship of ICL, IPL and RCL is illustrated in Fig. 1.5. The
The objectives of the consortium are to: ICL was briefly introduced in Landslides by Sassa
(2004); the full history of ICL was introduced in the pro-
1. promote landslide research for the benefit of society ceedings of the first General Assembly of the ICL held at
and the environment, and capacity building, includ- the U.S. National Academy of Sciences in Washington D.C.
ing education, notably in developing countries; on 1314 October 2005.
8 Kyoji Sassa

Fig. 1.5.
Cooperative structure of the ICL,
IPL and RCL for global landslide
risk reduction

and developing the International Programme on Land-

1.4 Development of the International Landslide
slides as a dynamic global network. Organizers of this
Community: the 2005 Letter of Intent,
discussion were the International Consortium on Land-
the 2006 Tokyo Action Plan, and
slides (ICL), United Nations Educational, Scientific and
the 2008 First World Landslide Forum
Cultural Organization (UNESCO), World Meteorological
Organization (WMO), Food and Agriculture Organiza-
The World Conference on Disaster Reduction (WCDR) tion of the United Nations (FAO), United Nations Inter-
was held on 1822 January in Kobe, Japan. At this confer- national Strategy for Disaster Risk Reduction (UN/ISDR),
ence, a session titled New International Initiatives for United Nations Environment Programme (UNEP), United
Research and Risk Mitigation of Floods (IFI) and Land- Nations University (UNU), Kyoto University (KU). It was
slides (IPL) was organized by the ICL, the United Na- cosponsored by Cabinet Office of Japan (CAO), Ministry
tions Educational, Scientific and Cultural Organization of Foreign Affairs, Japan (MOFA), Ministry of Education,
(UNESCO), the World Meteorological Organization (WMO), Culture, Sports, Science and Technology, Japan (MEXT),
the Food and Agriculture Organization of the United Ministry of Agriculture, Forestry and Fisheries of Japan
Nations (FAO), the Ministry of Education, Culture, Sports, (MAFF), Ministry of Land Infrastructure and Transport,
Science and Technology of the Government of Japan Japan (MLIT), Ministry of Foreign Affairs, Italy, Italian
(MEXT), the United Nations University (UNU), Kyoto Civil Protection Department (Presidency of the Council
University (KU), and others. Within this session, a Letter of Ministers), Ministry of Environment of the Slovak Re-
of Intent to promote further joint global activities in di- public, Ministry of Environment of the Czech Republic,
saster reduction and risk prevention through Strength- National Emergency Management Agency of Korea, Sci-
ening research and learning on Earth System Risk Analy- ence Council of Japan (SCJ), Japan International Coop-
sis and Sustainable Disaster Management within the eration Agency (JICA), International Union of Geologi-
framework of the United Nations International Strategy cal Sciences (IUGS), Academy of Forest, Wood and Envi-
for Disaster Risk Reduction (ISDR) by global partners: ronment, Japan (AFWE), and Japan Landslide Society (JLS).
UNESCO, WMO, FAO, UNU, ICSU, WFEO. The Letter of The Honorary Chairpersons for the round-table dis-
Intent can be an umbrella for all initiatives of Earth-sys- cussion were Salvano BRICENO (Director of UN/ISDR),
tem risk reduction. It was approved and signed by seven Hosny El-Lakany (Assistant Director-General of FAO),
global stakeholders, as shown in Fig. 1.6. Walter ERDELEN (Assistant Director-General of UNESCO),
Based on this Letter of Intent, a round-table discus- Michel JARRAUD (Secretary-General of WMO), and
sion was organized at the United Nations University in Kazuo OIKE (President of KU). The Chairpersons were
Tokyo on 1820 January 2006, in order to examine the glo- Hans van GINKEL (Rector of UNU), Yoshiaki KAWATA
bal plan promoting research and learning on landslides (Director of Disaster Prevention Research Institute of
Chapter 1 Landslide Science as a New Scientific Discipline 9

Fig. 1.6.
Letter of Intent strengthening
research and learning on Earth
System Risk Analysis and Sus-
tainable Disaster Management
within the framework of the
United Nations International
Strategy for Disaster Risk Re-
duction (ISDR)

Kyoto University), Badaoui ROUHBAN (Chief, Section for United Nations University, Tokyo, from 1820 January,
Disaster Reduction of UNESCO), and Kyoji SASSA (Presi- 2006 to formulate a framework for cooperation and to
dent of ICL, Director of Research Centre on Landslides identify focus areas to reduce landslide risk worldwide.
(RCL/DPRI-KU)). As the result of the discussion, partici- The action plan was adopted as a summary of the meet-
pants adopted the 2006 Tokyo Action Plan Strengthening ing, to be implemented within the scope of the Hyogo
Research and Learning on Landslides and Related Earth Framework for Action 20052015, Building the Resilience
System Disasters for Global Risk Preparedness. The ac- of Nations and Communities to Disasters, declared at the
tion plan is outlined from its preface: The 2006 Tokyo United Nations World Conference on Disaster Reduction
Round Table Discussion Strengthening Research and held in Kobe, Japan, in 2005. The full action plan is at-
Learning on Earth System Risk Analysis and Sustainable tached in the appendix of this volume.
Disaster Management within UN-ISDR as Regards Land- The Tokyo Action Plan proposed the World Landslide
slides-towards a dynamic global network of the Inter- Forum. Capitalizing on the competence, international
national Programme on Landslides (IPL) was held at the experience, and established organizational network of
10 Kyoji Sassa

ICL-IPL, this forum was proposed to create a global in- of the organizing committee were decided. The organiz-
formation platform for future joint activities of the ing committee and their members consist of all related
world-wide landslide community. The World Landslide United Nations Organizations, global stakeholders re-
Forum that shall convene every 3 years. The first World lated to this issue, landslide research groups and indi-
Landslide Forum is planned to take place in 2008, bring- vidual landslide experts. This first World Landslide Fo-
ing together academics, practitioners, politicians, et al., rum will assemble for four days and will extensively dis-
to a global, multidisciplinary, problem-focused platform. cuss the progress of research and learning to mitigate
The organization is in progress, and as of 15 December Earth system disasters, focusing on landslides. This fo-
2006, the dates, venue, organizers, and major members rum can be the second step for the global development

The First World Landslide Forum

Implementing the 2006 Tokyo Action Plan on the International Programme on Landslides:

Strengthening Research and Learning on Earth System Risk Analysis and Sustainable Disaster Management
within UN-ISDR as Regards Landslides
Date: 1821 November 2008, Venue: United Nations University, Tokyo, Japan

Organizers Geological Sciences (IUGS), Canada), Earl BRABB (President

Emeritus, International Landslide Research Group, USA), Ryu
International Consortium on Landslides (ICL), United Nations FUKUI (Manager, Tokyo Development Learning Center of the
Educational, Scientific and Cultural Organization (UNESCO), World Bank), See Sew GUE (World Federation of Engineering
World Meteorological Organization (WMO), Food and Agricul- Organizations (WFEO), Institution of Engineers, Malaysia), Tho-
ture Organization of the United Nations (FAO), United Nations mas HOFER (Forestry Officer, Specialist on watershed manage-
International Strategy for Disaster Risk Reduction (UN/ISDR), ment and mountain development, Forestry Department of FAO,
United Nations University (UNU), United Nations Environment Italy), Alik ISMAIL-ZADEH (Chair, Commission on Geophysical
Programme (UNEP), United Nations Development Programme Risk and Sustainability of International Union of Geodesy and
(UNDP), World Bank (IBRD), International Council for Science Geophysics (IUGG), Germany/Russia), Saroj Kumar JHA (Team
(ICSU), World Federation of Engineering Organizations (WFEO), Leader, Global Facility for Disaster Reduction and Recovery the
Kyoto University (KU), and Japan Landslide Society (JLS) World Bank Group, USA), Suzanne LACASSE (Managing Direc-
tor, Norwegian Geotechnical Institute (NGI)), Willy LACERDA
International Organizing Committee (Chair, Joint Technical Committee 1- Landslides and Engineered
Slopes, Brazil), Peter LYTTLE (Coordinator, Landslide Hazard Pro-
Honorary Chairpersons gramme, U.S. Geological Survey), Hideaki MARUI (President, Ja-
pan Landslide Society), Gordon MACBEAN (Chair, ICSU Planning
Salvano BRICENO (Director of UN/ISDR), Jacques DIOUF (Direc- Committee on Natural and Human-Induced Environmental
tor-General of FAO), Hans van GINKEL (Rector of UNU), Michel Hazards and Disasters, Canada), Norio OKADA, N. (President,
JARRAUD (Secretary-General of WMO), Kochiro MATSUURA (Di- Japan Society for Natural Disaster Science), Hari SRINIVAS (Chief,
rector-General of UNESCO), Goverdhan MEHTA (President of Urban Environmental Management, UNEP, Japan), Niria SANZ
ICSU), Shuzo NISHIMURA (Executive Vice President of Kyoto Uni- (Programme Specialist, World Heritage Centre of UNESCO,
versity) France), Yueping YIN (Director, Department of Environmental
Geology, China Geological Survey)
Local Organizing Committee
Edward BROMHEAD (University of London, Kings College, Lon-
don), Paolo CANUTI (European Centre of ICL, University of Flo- Chairperson
rence), Srikantha HERATH (Senior Academic Programme Officer
of UNU), Yoshiaki KAWATA (Director of Disaster Prevention Re- Kaoru TAKARA (Executive Director of ICL, Deputy Director of
search Institute of Kyoto University), Howard MOORE (Senior DPRI/KU)
Advisor of ICSU), Badaoui ROUHBAN (Chief, Section for Disaster
Reduction of UNESCO), Kyoji SASSA (President of ICL, IPL World Deputy Chairpersons
Centre), Robert SCHUSTER (U.S. Geological Survey)
Libor JANSKY (Senior Academic Programme Officer of UNU)
Members Hirotaka OCHIAI (Vice President of the Japan Landslide Society)

Robert F. ADLER (Senior Scientist, Goddard Space Flight Center Secretary General
of National Aeronautics and Space Administration (NASA), USA),
Peter BOBROWSKY (Secretary General, International Union of Hiroshi FUKUOKA (Treasurer of ICL, Associate Professor of DPRI/KU)
Chapter 1 Landslide Science as a New Scientific Discipline 11

of the landslide research community and the progress of Sassa K (1992) Access to the dynamics of landslides during earth-
as a new scientific discipline on landslides, namely Land- quakes by a new cyclic loading high-speed ring-shear ap-
paratus (keynote paper). In: 6th International Symposium on
slide Science. The already confirmed structure of the
Landslides, Landslides, A.A. Balkema. Christchurch, 3,
Forum is attached at the end of this article. Those who pp 19191937
are willing to contribute to the development of research Sassa K (1996) Prediction of earthquake induced landslides. In: Pro-
and learning on landslides and other Earth system di- ceedings of 7th International Symposium on Landslides, A.A.
sasters in the framework of the United Nations Interna- Balkema, Trondheim, 1721 June, 1, 115132
Sassa K (2000) Mechanism of flows in granular soils. In: Proceed-
tional Strategy for Disaster Risk Reduction (UN-ISDR)
ings of the International Conference of Geotechnical and Geo-
are requested to join this forum. Any groups or any or- logical Engineering, GEOENG2000, Melbourne, 1, pp 16711702
ganizations are invited to propose and organize various Sassa K (2004) The International Consortium on Landslides. Land-
types of session, workshops, or symposia to promote re- slides 1(1):9194
search and learning on landslides and other related Sassa K (2005) ICL history and activities. In: Sassa K, Fukuoka H.,
Earth-system disasters in this forum. Rooms and facili- Wang FW, Wang G (eds) Landslides Risk analysis and sustain-
able disaster management. Springer-Verlag, Berlin Heidelberg,
ties will be organized by the organizing committee. pp 321
Sassa K (2006) 2006 Tokyo Action Plan-strengthening research and
leaning on landslides and related earth system disasters for glo-
References bal risk preparedness. Landslides 3(4):361369
Sassa K, Fukuoka H, Wang G, Ishikawa N (2004) Undrained dynamic
Antonello G, Casagli N, Farina P, Leva D, Nico G, Sieber AJ, Tarchi D loading ring-shear apparatus and its application to landslide
(2004) Ground-based SAR interferometry for monitoring mass dynamics. Landslides, 1 (1), pp 719
movements. Landslides 1(1):2128 Varnes DJ (1958) Landslide types and processes. In: Eckel EB (ed)
Canuti P (2004) ICL 2003 Varnes Medal. Landslides 1(2):163164 Landslides and engineering practice. Highway Research Board,
Cruden DM (1991) Cowboys and Landslides. Landslide News 5:3132 Special Report 29, pp 2047
Cruden DM, Varnes DJ (1996) Landslide types and processes. In: Varnes DJ (1978) Slope movement types and processes. In: Schuster
Turner AK, Schuster RL (eds) Landslides: investigation and miti- RL, Krizek RJ (eds) Special report 176: Landslides: analysis and
gation. Special Report 247, Transportation research board, US control. Transportation research board, National research coun-
National research council, Washington, D.C., pp 3675 cil, Washington, D.C., pp 1133
Chapter 2

An Overview of Landslide Problems in the British Isles, with

Reference to Geology, Geography and Conservation

Edward N. Bromhead* Maia L. Ibsen

Abstract. An overview is presented of the landslide problems expe- The native inhabitants of Britain think that their is-
rienced in Britain, and their primary causes. Principally, landslides lands are generally wet, and that this is the main factor
occur in Britain where the strata are argillaceous and there is suffi-
governing slope instability, but the appearance of wetness
cient topographic relief. This combination occurs in swathes through
the southern and central parts of Britain and along the lengthy comes not from the total rainfall, but from its generally
coastline. In these parts of Britain, the main rock types are sedimen- all year round occurrence, and the absence of a long dry
tary, and they often exhibit low angles of dip. This gives rise to the season to generate a build up of soil moisture deficit. In-
occurrence of compound landslides, often with a bedding-con- deed, although changes in rainfall are a major factor in
trolled flat basal shear surface. Issues relating to this type of land-
activation and reactivation of landslides, a survey of land-
slide are discussed in the article, which concludes with a discussion
of the conflicts between interests of land users and the need for slides in the literature revealed (Jones and Lee 1994; DOE/
conservation, especially in some areas that are of significant inter- GSL 1987) that the majority of the recorded instances were
est in the history of the geological and other sciences, and to re- associated with erosion along the coastline (especially
search in the present day. where argillaceous rocks and soils formed sea cliffs of
moderate height); along the outcrops of clays of Meso-
Keywords. Bedding, slip-surface, landslide distribution
zoic and Tertiary age, which sweep across country in sev-
eral well-defined bands (Fig. 2.1); in the Coal Measures (Car-
boniferous); and with a few areas where glacial erosion, and
2.1 Introduction to a lesser extent, river erosion have oversteepened inland
valleys again, in the clay strata, and where glacial materi-
The British Isles contains geomaterials from most of als have been cut into by stream or coastal erosion.
the geological record, and is one of the locations where A further factor affecting the coast is rising sea level
major steps were taken in unravelling earth history in during the past few millennia, and the consequent increase
the early years of the geological sciences. This is evident in erosion rates, so that with only a comparatively few
in the many names which emanate from the British Isles exceptions, the most active landslides are found on the
within the geological time scale: these names range from coast where the slide-prone strata are intercepted by the
the names of tribes resident in Britain 2 millennia ago evolution of the coastline. Inland, several generations of
(Ordovician, Silurian) to the names of particular locali- transportation technologies (roads, canals, railways and
ties (e.g., Devonian, London Clay), and in this respect, roads again) have cut swathes through the landscape, and
the country is grossly overrepresented in geological these have generated first-time failures in both cut and
nomenclature. In general terms, the rocks are oldest in fill slopes (some of which merit designation as engi-
the north of Scotland, and in the west, although episodes neered slopes) and have reactivated ancient slides, again,
of glaciation have emplaced a variety of soils of glacial where they cross the outcrops of the most slide prone
origin to add interest to the landscape by putting an strata. Failures in engineered slopes either involve the
irregular veneer of weak soils over the oldest and stron- reactivation of fossil (usually periglacial) landslides, or
gest rocks. The southern half of the main island (England, are generally similar to first time failures in natural slopes,
Scotland and Wales) of the British Isles is composed of so the inclination of British practitioners is not to rigor-
weak Mesozoic and Tertiary sedimentary rocks, and ously separate natural slopes from man made ones, and
these usually give rise to the most interesting of the that approach is followed here.
landslides. Landslides are not confined to particular types of
geomaterials, but given a steep enough and high enough
slope, together with an input of water and in some cases,
* The corresponding author of each chapter is marked with an ground accelerations due to seismicity (uncommon, but
asterisk. not completely unknown in Britain), landslides of vari-
14 Edward N. Bromhead Maia L. Ibsen

Fig. 2.1.
Sketch map of Britain, showing
principal geological units and
the landslides found in a survey
of the literature (Jones and Lee
1994; DOE/GSL 1987). Main
locations mentioned in the text
are also shown

ous types occur in every rock type. As a result, the usual

2.2 Strong Rocks and
initial classification of geomaterials on the basis of their
Discontinuity-controlled Slope Instability
origin (igneous, sedimentary, metamorphic) is less use
than it might at first seem.
The geotechnical engineer usually finds it convenient Vulcanism in Britain is ancient, and there are no loose
to divide geomaterials into rocks and soils largely on pyroclastic deposits susceptible to slide or flow mecha-
the basis of the ease with which they can be excavated, nisms, so that in Britain igneous rocks are usually seen
and developments in the power of excavation plant have as strong rocks. Most igneous rocks contain joints and
made this junction rather indistinct. The landslide spe- other discontinuities resulting from cooling and subse-
cialist may well consider a classification based on whether quent regional tectonic stresses. The resulting joints have
or not stability is conditional principally on failure close to moderate spacing, resulting in comparatively
through pre-existing discontinuities or through the mass small block sizes when falls occur. A good example of
of material between discontinuities as respectively indi- this is the Whin Sill (Fig. 2.2) of columnar-jointed eclog-
cating that the material is a rock or a soil. ites intruded into limestones of Carboniferous age, and
Chapter 2 An Overview of Landslide Problems in the British Isles, with Reference to Geology, Geography and Conservation 15

Fig. 2.2.
The Whin Sill is formed from
rocks with columnar jointing,
and exhibits scree slopes, al-
though the rock blocks have
been exploited for building for
two millennia, not least in the
construction of a defensive wall
in the 2nd century AD which still
remains (albeit in ruins) at the
crest of the cliff

Fig. 2.3.
Bedding control of the basal shear
or slip surface of the Taren land-
slide forms a staircase, follow-
ing a downfaulted seatearth in
the Coal Measures (Kelly and
Martin 1985)

forming a pronounced outcrop across northern England.

2.3 Weak Rocks and Strong Soils
This was exploited by the Roman Emperor Hadrian for
part of his wall (2nd century AD) separating the barbar-
ians to the north from the civilized south. Angular frag- Glaciation appears to have removed any areas of deep
ments of rock, falling few at a time, tend to accumulate weathering, and so problems of, for example, the appear-
in scree slopes, as seen in the photograph, and provided ance of clays in weathered granites such as plague Hong
that care is taken to avoid the areas of highest impact Kong are not experienced. However, materials of glacial
risk, generally pose little problem. origin are found in most of the upland, and some low-
Occasionally, Tertiary lavas have capped some depos- land, areas in northern and central Britain, and these prove
its of Mesozoic rocks (e.g., the Jurassic clays of the Lias to be susceptible to being involved in debris flows. Natu-
in the Isle of Skye forming the Trotternish landslides), rally-occurring large debris flows are unusual, although
so that the resulting landslides are compound block slides failures in mine waste tips in coal mine and china clay
with strong caprocks, but these are few and far between. waste have provided examples, including the Aberfan
What passes for a mountain even in the highest parts mine waste tip landslide in October 1966 that generated
of Britain would be seen as a hill in many other parts of by far the UKs highest toll of casualties 144 as fatal-
the world, and the relief is not conducive to the for- ity-inducing landslides are not unknown, but tend to cause
mation of large landslides in the stronger rock types. In the loss of one or two lives per incident.
consequence, large landslides in Britain are often slow Many of the mudrocks of the late Paleozoic, Mesozoic
moving. and the Tertiary are comparatively weak, and fall in that
16 Edward N. Bromhead Maia L. Ibsen

transitional group of weak rocks or strong soils. Per- slides with a step-form of basal slip surface are found.
haps the oldest of these in Britain are the Coal Measures Perhaps the best known example of this from the litera-
(Carboniferous), best known from its outcrop in the Welsh ture is the Taren landslide (Kelly and Martin 1985; Fig. 2.3)
Coalfield, an upland syncline in south Wales cut through reanalysed in 3D by Bromhead and Martin (2004).
by numerous valleys, and therefore exhibiting landslide Similar staircase forms are found in many slopes, not
after landslide in essentially the same strata on both sides generated by down-faulting of the critical beds, but where
of, and throughout the length of, each valley (Siddle et al. slipping drapes shallow mudslide debris over a slope with
2000). The downfaulting present in many of the valleys a series of more competent beds, for example, as found in
leads to the weak beds forming a staircase, so that land- the East Cliffs at Lyme Regis (Clark et al. 2000). These

Fig. 2.4. Landsliding in the coastal cliffs of Barton on Sea (Barton et al. 2006). Several weak zones in the mainly clay sequence are exploited
for basal shear surfaces of the slides, most of which are perched, i.e. daylight in the cliff. This length of coastal slope has been exempted
from slope stabilization in view of its geological interest

Fig. 2.5. A major bedding controlled landslide at St. Catherines Point, Isle of Wight (Hutchinson et al. 1991). The slip surface is located in a
thin clay layer, which is below sea level. Geological sequence is primarily mid to lower Cretaceous in age. Dips are seaward
Chapter 2 An Overview of Landslide Problems in the British Isles, with Reference to Geology, Geography and Conservation 17

staircase sequences are very common, with numerous especially for railways (which are often steeper and deeper
small steps, or fewer large steps. Examples with fewer steps than the corresponding road cuts). Examples about in the
are seen in the Barton Beds (Barton 1977, 1984; Fig. 2.4), literature, from as long ago as the mid nineteenth century
and in the SW coast of the Isle of Wight (Hutchinson et al. (Gregory 1844), through the golden era of London Clay
1985), although in these cases the more competent beds and Imperial College slopes research (Henkel 1957) right
are weak sandstones, often of locked sand and not the through to modern times (Kovacevic et al. 2003).
thin muddy limestones of the Lias.
Such landslides that owe their overall morphology to
one or a few slide prone horizons in a complicated se- 2.4 Erosion
quence of beds are commonly termed bedding-con-
trolled slides (Bromhead and Ibsen 2002). In exceptional The dominant form of erosion experienced in the UK is
cases, the staircase landform and matching basal slip sur- coastal erosion. Sea level has been both slightly higher,
face is involved in a massive landslide involving displace- and substantially lower, in the comparatively recent past.
ment often a deeper bedding surface, usually through a Raised beaches at Black Rock, Brighton, and on the Isle of
plastic clay layer, as in the case of the St. Catherines Point Portland, attest to a sea level 6 or 7 m higher during a
landslide on the Isle of Wight (Hutchinson et al. 1991; particular time in the Pleistocene than at present, and sea
Bromhead and Ibsen 2002 see Fig. 2.5), although equally, levels have been much lower in the historic past. For ex-
there are instances of such landslides occurring not as a ample, the maps of the Italian Coastline by Margottini
few blocks, but as many successive more rotational slides, and Vai (2004) show a sea level 100 m lower than present
as at Folkestone Warren (Hutchinson et al. 1980). How- as recently as 22 2 ka BP, and those few meters higher
ever, bedding-control of the basal shear is a well-estab- around 8 1 ka BP. Hutchinson and Bromhead (2002) re-
lished phenomenon. Indeed, the Skye landslides are best produce a sea level rise curve which shows 15 m of rise in
understood as bedding-controlled and thus compound the past 8 millennia. There is conflicting evidence of pre-
landslides with a massive caprock. cise levels, positions of the coastline, and dates, but the
Naturally, the presence of caprocks permit the mul- overall picture is one of the sea level lowering and aban-
tiple block nature of many of these deep seated bedding doning coastal cliffs, which were then able to degrade sub-
controlled landslides to be more readily understood than aerially with extensive aprons of slide debris at the toe,
where they occur in single lithologies, although the same before being caught up in rising sea level and the buried
general processes occur in the coastal London Clay cliffs of sea cliff being re-exhumed from underneath its covering
North Kent (Bromhead 1978; Dixon and Bromhead 2002). of landslides. Such coastal slopes are far from being in
Bedding controlled basal shear locations are not con- equilibrium, and are thus subject to rapidly worsening
fined to natural slopes, but occur commonly in cuttings, stability as the slope toes are eroded.

Fig. 2.6.
The worlds first cast-iron bridge
across the Ironbridge Gorge
(a modern name) is threatened
by slope movements from both
sides of the gorge. A concrete
structure has already been built
in the riverbed
18 Edward N. Bromhead Maia L. Ibsen

Inland, glacial erosion has oversteepened slopes in such as cliff retreat or the movement of built-up areas,
many upland areas, and combined with stream or river detailed studies of many of the landslides have been un-
erosion, in some cases by streams rejuvenated by lowered dertaken. Such studies have revealed the control exerted
sea levels, has had further effects. In the Ironbridge Gorge, on landslide morphology by geological structure. Often,
where the river Severn leaves a formerly ice-dammed lake bedding has a low angle of dip, and such geological con-
(only the river meander across the sediments of the lake ditions usually give rise to a graben like structure or mul-
bottom attest to its former extent) to pass through a gla- tiple/successive rotational slips. Successive failures lead
cial lake erosion channel, the resulting cut through Coal to a regular, stepped pattern in the landscape. A graben is
Measures (which is still eroding at the present time) has the zone of subsidence adjacent to the detachment sur-
led to landsliding on both sides of the valley. This gorge face of a landslide. As well as subsidence, there may also
exposed all the necessary minerals for iron smelting (iron be block rotation and disruption in this zone.
ore, limestone, low-sulfur coal and fireclay for refracto- Studies of aspects of the groundwater behavior, in par-
ries) in close juxtaposition, and the area became the fo- ticular its response to weather and to erosion rate, have
cus for the industrial revolutions. Inevitably, the stability provided deep insights into the mechanics of failure and
of the area was further compromised by underground of post-failure movement. However, the geological struc-
workings and surface spoil heaps, as is everywhere that ture provides the primary control on slip surface posi-
the Coal Measures have been mined underground. Move- tion. Slides may take place where the beds dip strongly
ment of the as-yet incompletely understood landslides out of the slope, i.e. coastward or valleyward, or where
threatens the historic Ironbridge (Fig. 2.6), the worlds first the dip is gently out of the slope. Paradoxically, they can
cast-iron bridge. (This bridge was assembled with wedges also occur where the dip is into the slope, but only when
and dovetail joints, because it predated the regular appli- the dip is comparatively gentle, since a steep dip into the
cation of bolts or rivets in structural work!). Erosion of slope inhibits sliding. The loose terms steep and gentle
these types and slope movements pose problems in the when applied to dip are relative to the mechanical prop-
conservation of historic sites and monuments, although erties of the soils and rocks involved. For the purposes of
equally and paradoxically, some locations are of high cul- this paper, the boundary between the two occurs at a dip
tural value due to erosion. Perhaps the best example of equal to approximately half of the residual angle of shear-
this is the designation of the Dorset Coast as a World ing resistance. In addition, a very steep dip is above about
Heritage site, on account of its historic role in the devel- half of the peak angle of shearing resistance.
opment of the geological sciences. A major contribution to landslide science from Brit-
ain is related to the understanding of the interplay be-
tween geological structures and the internals of the land-
2.5 The Coastline of South Eastern Britain slide systems that result, together with an understanding
of the morphology of the topographic surface. The prin-
The coastline of Southeast Britain is formed in sedimen- cipal factor is the existence of a low-angle clay bed in
tary rocks of Jurassic, Cretaceous and Tertiary age, the ma- which the main basal shear surface can form. In one or
jority of these deposits contain thick strata of mudrocks, two rather exceptional cases, this follows the line of an
which have very low angles of dip. Where these strata are unconformity. There are therefore numerous similarities
appropriately exposed, particularly at the foot of a coastal between sites that have been studied, and broad correla-
slope, they give rise to landslides where all or part of the tions may be identified. However, the precise mechanics of
sliding surface follows a single bed of mudrock. Where each landslide are different, affected by local dips, folding,
the dip of the bedding is steep, such landslides are re- faulting and lithologies. Indeed, the position within the
ferred to as dip-slope failures, and they are usually trans- coastal cliff at which the clay beds are found, has an im-
lational. However, landslides where the basal sliding sur- portant role in governing, inter alia, the response to toe
face is controlled by the location and orientation of a single erosion, itself a variable due to the differing degrees of
argillaceous bed in the sequence are better termed bed- exposure to marine action around the coastline. Some of
ding-controlled landslides and they are usually com- these differences are drawn out in the following sections.
pound slides. In addition, where coastal slopes contain sev-
eral layers of mudrocks, geometrically similar landslides
may occur with perched slide surfaces breaking out at a 2.6 Bedding-Control of Slip Surfaces
higher level within the slopes. Landslides with strong bed-
ding-controlled basal shear surfaces are the predominant The development of the slip circle by Petterson (1955),
form of instability along the Southeast coastline of Britain. although a great advance in the understanding of the
In this part of Britain there are numerous coastal towns, mechanics of slope failure, locked engineers into a
which are on, or near, these active landslides. As a conse- mindset which ignores the role of geological structure on
quence of the various threats posed by these landslides, slope failure. Petterson himself merely simplified the ge-
Chapter 2 An Overview of Landslide Problems in the British Isles, with Reference to Geology, Geography and Conservation 19

ometry of rotational slope failures after Collin (1856)

whose elaborate spiral shape curve-fitting exercises ig-
nore the geological control which is evidently present even
in thick comparatively uniform strata of mudrocks. By
contrast, Gregory (1844) clearly indicates the presence of
a slip surface aligned to bedding in his failed slope sec-
tion even if it is not referred to explicitly in his text.
Terzaghi and Peck (1948) classified slip surface posi-
tions as slope failure, toe failure and base failure
(Fig. 2.7a) where the emergent slip surface position is re-
spectively on the slope face, precisely at the toe of the slope,
or where the slip surface is sufficiently deep-seated to
emerge beyond the toe of the slope. His sketch was drawn
in terms of slip circles, and a redrawing of the figure to
include bedding-controlled landslides creates the variant
shown in Fig. 2.7b according to Ibsen and Bromhead (1999). Fig. 2.7. Toe breakout position in the slope for different types of land-
slide. Terminology in (a) by Terzaghi and Peck (1948), and (b) by
Note that in this figure the dip of the beds is zero, Similar Ibsen and Bromhead (1999)
diagrams could be drawn with non-zero dips, both in a sea-
ward sense and in a landward one (for a coastal slope).
Reasons cited in an extensive research of the literature 2.7 Slides in Strata with Low Angle Dips
on these landslides for the occurrence of bedding-con-
trolled slide surfaces include: Landslides where the bedding surfaces follow strata with
low angles of dip are common in Southeast Britain, not
a zones of high plasticity or liquid limit in which the slide merely on the coast. A cross section of a slope failure in a
surface forms; railway cutting is given by Gregory (1844) of a landslide
b change in rigidity above a hard band, causing local- that occurred in 1841 in a railway cutting at New Cross in
ized strains to occur under stress relief; south London. He describes the slip surface as glass like,
c perched water table conditions resulting from a change and clearly shows it following the bedding (Fig. 2.8), which
in permeability; here is sub-horizontal and is clearly recognisable as it is
d basal erosion cutting down to a limiting horizon (hard the transition between the weathered and unweathered
band). beds of the London Clay. Much more recently, Cooper et al.
(1998) created a small landslide in Gault Clay in a pit dug
Various authors have speculated that slip surfaces for brick-making at Selborne. Their failure (Figs. 2.9 and
develop in the unsheared ground at or about the base of 2.10) followed the bedding a little above the toe of the
a cut slope, thus providing a pre-existing slip surface to slope, and the slip surface could also fairly be described
be followed by a major failure. Attempts to model or as glass like if one refers to the reflectivity and polish (not
detect such a developing surface have been made on transparency).
numerous occasions, with varying degrees of success. However, failures in cut slopes are likely to be cleared
Such a feature has been found in practice by Burland et al. away during remedial works. Failures in natural slopes,
(1977), and by Cooper et al. (1998). Both of these slopes especially along the coast, are rather more permanent fea-
were in pits where overconsolidated clays were extracted tures of the landscape.
for brick making, but the principle is otherwise the same. At Red Cliff (Bromhead and Ibsen 2003), Folkestone
In no natural slope or coastal landslide case has the (Hutchinson et al. 1980) and Herne Bay (Bromhead 1978),
necessary field instrumentation been installed or moni- the basal slip surface locations are close to the base of the
tored, although modeling by both limit equilibrium and clay stratum in which they have formed. This has given
finite element methods is strongly indicative of such a rise to speculation that the slip surface has formed where
development. the differential shear strains are concentrated. In such a
The evidence as to whether such a basal slide surface case, the basal shear surface location could result from
is the result of a specific lithology in the geological se- stress-relief related expansion of the slope following ero-
quence is mixed, with strong support only at Folkestone, sion. At other locations, notably Gore Cliff, Bromhead et al.
where the slip surface location of the Folkestone Warren (1991), and the Roughs, Hythe, Bromhead et al. (1998),
landslide complex is clearly associated with a band of high the critical slip surface location lies, as at Selborne, in the
liquid limit clay. No doubt even extremely subtle litho- middle of a thick stratum and is not easily correlated with
logical differences may provide the necessary conditions any major lithological change. Chandler (1984) does im-
for slip surface formation. plicate a change in plasticity for the location of the slip
20 Edward N. Bromhead Maia L. Ibsen

surface at Gore Cliff. The dip at Red Cliff, as in a number a long enough period, however, coastal retreat will en-
of other coastal localities, is into the slope. For the low counter significantly different geological structures, with
angles of dip present over most of the area, such a dip a consequent change in landslide activity and the rate of
direction does not significantly inhibit slide development, coastal retreat.
although it clearly has an effect in stability analysis, since For landslides primarily in debris, it has been specu-
where the slide behavior is dominated by its basal shear, lated that the process of basal incorporation, Hutchinson
each 1 of dip adds or subtracts the effect of approximately (1972), may occur until numerous mudslides or debris
1 to the mobilized angle of shearing resistance '. Over slides scour down to a common basal level. This is in-
consistent with some of the mechanisms which pre-shear
a particular horizon. Basal incorporation appears to be
a major factor with large and active mudslide systems,
and these appear to be able to gouge down through hard
bands in the sequence, and easily cut through pre-exist-
ing bedding-controlled basal shear zones. Underneath
existing landslide accumulations, particularly when they
are inactive for long periods, the processes of softening
and weathering may continue. The slip surface is not a
significant barrier to this and it is not surprising, there-
fore, that the basal shears of bedding-controlled land-
slides are rather easier to find in active landslides than
in quiescent ones, or in some abandoned slopes. This may
Fig. 2.8. Slide in a railway cutting at New Cross (after Gregory 1844).
appear to contradict the observation of basal erosion or
The basal shear surface was described as glass-like. It follows a bed- basal incorporation, where the materials underneath the
ding surface at the base of the London Clay formation slide surface of a mudslide are softened and sheared out

Fig. 2.9.
Slope failure at Selborne in-
duced by groundwater recharge.
The slip surface breaks out above
the toe of the slope

Fig. 2.10.
Cross section of the failure in-
duced in the Selborne pit show-
ing correlation of slip surface
position with the bedding,
although the weak bed here
(sketched in as a white band)
was not detected in the investi-
gation or instrumentation stages
in the project
Chapter 2 An Overview of Landslide Problems in the British Isles, with Reference to Geology, Geography and Conservation 21

as the mudslide develops. Such basal erosion, scouring This situation is illustrated in Fig. 2.11, which is a
down to a common horizon in many, closely adjacent or sketch cross of the section of the Red Cliff landslide. In
superimposed mudslides, might be the reason for the the upper part of the figure, both the main landslide
occurrence of bedding-controlled landslides where fur- scarp and the antithetic scarp break out in flat ground
ther penetration is inhibited by a strong horizon (e.g., a at the top of the slope. In this case, the resulting ridge
limestone in the Lias), but this is improbable where any in the slide mass after movement is at, or close, to the
thickness of mudrock separates the strong bed from the elevation of the crest of the pre-slide slope. An im-
slip surface. Indeed, mudslide cascades appear capable pression of decrease in level is gained, however, when
of cutting down through virtually any geology if they the main slide movement is down-dip. Alternatively,
are active enough. the breakout position of the antithetic scarp may be in
the main pre-slide slope. This leads to a somewhat dif-
ferent morphology, particularly where the slope has
2.8 Grabens and Graben Geometry a thick, strong, caprock (Fig. 2.5 the St. Catherines Point
landslide). The graben in such a case, if it may still be
Bedding-controlled slip surfaces, especially where dips called a graben, is a separately back-tilted mass of soil
are low, give rise to the development of grabens at the and rock.
head of the slide. A part of the slide mass moves largely Cruden et al. (1991) suggest a correlation between
in a downwards direction and may rotate slightly, how- graben depth and width and Ibsen and Bromhead (1999)
ever, the bulk of the slide mass travels in a different have extended this correlation for coastal landslides in
direction, along the line of the bedding, and there has Southeast Britain. A limited amount of experience sug-
to be an accommodation displacement between these gests that the head-to-toe distance is longer where the
two major parts of the landslide. This often takes the dip is out of the slope (seaward) than when it is into the
form of a shear surface, and thus a counterscarp or slope (landward). Furthermore, the latter case also tends
antithetic scarp (Cruden et al. 1991) appears in the sur- to push the counterscarp towards the toe, and therefore
face morphology. away from any flat ground at the head of the slope.
Grabens in landslides take their characteristic mor-
phology (Figs. 2.12 and 2.13) from the changes in
direction in the slip surface shape, and the abruptness
or otherwise of those changes. The model for graben
geometry developed by Cruden et al. (1991) stated
that it was possible to estimate the depth of the slip
surface, from the original ground surface, using the ini-
tial graben width, based on observations of ten small
landslides in Alberta, Canada. They indicated that the
depth to the translational slip surface was approximately
1.1 times the original graben width. Ibsen and Bromhead
Fig. 2.11. Sketch cross section of the graben-type compound land- (1999) have extended this field relationship between
slide at Red Cliff, Dorset graben width and depth to slip surface, with additional

Fig. 2.12.
Perched slip surfaces result from
a weak bed that daylights in the
face of a slope
22 Edward N. Bromhead Maia L. Ibsen

The control exerted by bedding on slip surface loca-

tion is very clearly seen where the slip surfaces are perched
in the slope. Perched systems of landslides are related to
the occurrence of a weak bed or beds (usually clays) in
the geological sequence and located above beach level. In
these cases, the landslide morphology follows the geologi-
cal structure, often depicting a stepped pattern.
In the simplest cases, the sequence contains only one
weak bed, and a single perched slide appears in the slopes.
Even where the elevation of the weak bed is only a little
above beach level and the sea cliff is therefore low, any debris
which spills over the sea cliff from slide activity at a higher
level is readily removed, thus keeping the sequence clean.
Even if debris is not removed, then it takes the form of
mudslides and screes (depending on the nature of materi-
als present and their water content), which are easily recog-
nised as different from the higher-level slide morphology.
Single perched slides may also occur in abandoned or
defended cliffs. For example, the slopes at The Roughs,
Hythe, Bromhead et al. (1998) or at Hadleigh Castle,
Hutchinson and Gostelow (1976). In such cases, the el-
evation of the basal slip surface may be obscured by the
build-up of slide debris and vegetation. The slopes un-
derneath the rotational slide may be occupied by a trans-
port zone where mudsliding occurs, predominantly par-
allel to the terrain slope, and an accumulation zone. The
slides may also occur with bedding dipping into or out of
the slope, or indeed, with a component of dip obliquely to
the trend of the coast line.
Fig. 2.13. Evolution of a graben in a compound landslide Where there are two or more weak beds, they may be
widely spaced, in which case separate bench-like systems
data and refining the relationship using further regres- of the landslide may be found. Alternately, they may be
sion analysis techniques. closely spaced, and then the debris from the uppermost
slide system may build up on the head of a lower system,
and thus obscure the position of the basal shear of the
2.9 Perched Slip Surfaces upper system. This is most likely where the slope is not
subject to active erosion at its toe, because the process of
Where clay beds outcrop above the overall toe of a slope, undrained loading (Hutchinson and Bhandari 1971) leads
one or more perched stratigraphically-controlled benches to increased activity in the lower slopes and re-exposure
(Fig. 2.12) may appear in the slope. The control exerted of the upper slide break-out position.
by the bedding on the slip surface location is very clearly In some cases, it is inferred that there are two weak
seen in this case when the critical bedding surfaces are layers in a single geological unit, so that the slip surface is
widely spaced and the rate of erosion is such as to pre- stepped. Where borehole data is used, and the boreholes
vent the accumulation of debris. In these cases, the land- are widely spaced, it may appear that the basal slide surface
slide morphology follows the geological structure, often is unrelated to bedding. This is particularly marked where
depicting a stepped pattern. the bedding dips into the slope. For example, at the Roughs,
Although hillslopes are low angled in most of Britain, Hythe, the regional dips have a component inland.
slope heights of 30 m or more are common, and these are
sufficient to give rise to deep-seated landslides where coast
erosion attacks the foot of those slopes. Hillslopes rising 2.10 Slip Surfaces at or close to the Base of a Slope
to over 100 m in height may be the location of larger-scale
landslides, particularly where the clay beds outcrop to- Slip surfaces of some coastal landslides break out in the
wards the foot of the slope. Alternatively, where critical beach (base failure or over-ridden toe position) or at
clay beds outcrop higher in the slope, one or more perched the junction of the erosion platform and slope (toe failure
stratigraphically-controlled benches may appear. or runout toe position). Classic examples are the major
Chapter 2 An Overview of Landslide Problems in the British Isles, with Reference to Geology, Geography and Conservation 23

landslides at Sheppey and Herne Bay in the London Clay,

Dixon and Bromhead (1991, 2002), Bromhead (1978); in
the Chalk, Upper Greensand and Gault of Folkestone
Warren, Hutchinson et al. (1980), Trenter and Warren
(1996); and in the Ventnor Undercliff, e.g., at St. Catherines
Point, Hutchinson et al. (1991), where the Sandrock se-
quence of the Lower Greensand contains the basal shears.
Folkestone Warren (Hutchinson et al. 1980) occupies
the entire length of coastal cliff where the Gault outcrops.
At its western end, the base of the Gault emerges above
the shore platform. The Warren landslides extend to the
east where the Gault disappears beneath sea level. Trenter
and Warren (1996) report movement patterns which are
greater to the west, where there is less passive restraint to
the breakout of the toe of the landslides. These slides,
therefore, appear to be rotating in plan. Also, the Warren
becomes generally wider towards the west possibly as a
consequence of this increased activity.
The Herne Bay landslides, Bromhead (1978), show an-
other interesting facet of behavior. The three major land-
slide remnants encountered by slope stabilization works
are more clearly bedding-controlled where the junction
Fig. 2.14. Scars in the foreshore from previous landslides at Warden
with the underlying Oldhaven Beds is close to sea level Point, Isle of Sheppey (Dixon and Bromhead 2002). The basal slip
(Miramar landslide), and less so where the junction lies surfaces are below sea level, and ongoing coastal recession planes
deep below sea level (Beacon Hill landslide). The Miramar off the slide masses, leaving these scars in the beach
landslide was a first time failure in 1953, and contempo-
rary photographs show a clear graben. In contrast to Many of the smaller landslides are primarily slides of
Folkestone Warren, the Miramar landslide was more ac- debris, with the original in-situ geological succession
tive at its western end, where the slip surfaces were below above the basal shear unrecognisable except in small
sea level, although their depth was never sufficient to cre- blocks. This contrasts with the larger landslides, where
ate a major passive resistance, and marine erosion was massive blocks of landslipped material may retain its origi-
able to make a significant impact on what passive restraint nal appearance and succession. Where the slides are of de-
there was, until the site was stabilized in 1969. bris, and are perched in the cliff, the basal shear surface
At Warden Point, Isle of Sheppey, Kent (Dixon and outcrop may be marked by seepages, by a color change, or
Bromhead 2002), the basal slip surface position lies be- by a step as the debris moves forward en masse. The latter
low sea level, and the rapid coastal retreat there has left feature appears from time to time, and will be recognised
scars of planed off landslides in the foreshore, where in the field or in terrestrial photographs (Fig. 2.15 a bed-
they are clearly visible at low tide (Fig. 2.14). ding-controlled basal shear of a compound landslide in
the rapidly-eroding cliffs at Fairlight in Sussex). It is rarely
if ever large enough to be identified in an air photograph
2.11 Identification of Slip Surface Position (although conceivably it could be if a shadow was cast)
since sufficient forward projection to be seen would re-
Hutchinson (1983) discusses the problems associated with sult in the collapse of the toe of the slide mass which be-
locating slip surfaces in pits, boreholes and soil samples. ing composed of debris is weak and to a certain extent
However, the breakout of a bedding-controlled slip sur- lacks cohesion.
face may be evident from the morphology of the slope alone. A major problem with studies of these coastal land-
For example, the location of bedding-controlled sliding sur- slides is that it is difficult to view them from a suitable
faces perched in a slope in an active landslide system can angle to understand their morphology. Stereoscopic ver-
be identified visually. Clearly, for slopes covered in vegeta- tical air photographs can help, but often need a trained
tion, or greatly degraded and covered in debris, this is not eye to interpret them. However, low-angle oblique air
possible. Furthermore, since the outcrop of the line of the photographs clearly reveal the details of the overall mor-
slip surface is on or near a vertical face, it may be identi- phology and in many cases, the location of the breakout
fied most easily visually in the field, or on photographs. position of the bedding-controlled slip surfaces can be
Oblique air photographs are best: vertical photographs identified. The depth of slip surface in the slide mass may
are of little use in showing the detail on vertical faces. also be estimated from the graben geometry.
24 Edward N. Bromhead Maia L. Ibsen

2.13 Conservation and Conflict

In the late nineteenth and early twentieth century, it

became the practice in Britain to protect vulnerable
coastal sites from erosion and inundation. Such sites
included both low-lying ground and some sites where
coastal slopes and cliffs of weak rocks are susceptible to
landsliding. In the latter case, cliff retreat rates in south-
eastern England can be of the order of a meter per year.
It has probably never been economic to protect agri-
cultural land, nor rural settlements, and whatever form
of cost-benefit analysis is employed (Lee and Jones 2004),
coastal protection and stabilization schemes are only
Fig. 2.15. Bedding-controlled slip surface in the Fairlight Clays meaningful where the built environment has a sig-
(Sussex) nificant value, for example, a large village or small town
is the minimum that generates the benefit to counter
the major cost of building the works in the first instance,
2.12 Three Dimensions and the Plan Shape and of maintaining them thereafter. However, some
of Landslides works have been constructed to protect land of mar-
ginal value, possibly for political rather than economic
Bromhead et al. (2002) describe a small coastal landslide reasons, although this practice is less prevalent than some
of the bedding-controlled type at Hanover Point, Isle of decades ago. Since it is difficult to separate the effects
Wight. This landslide occurs in mudrocks of Wealden age. of coast erosion from those of the movement of coastal
At the site of the landslide, they dip inland, and also have landslides, the costs of the former include the stabiliza-
a coast-wise component of dip. The morphology of the tion of the latter.
landslide is seen to have a dip into the slope as erosion Towards the end of the twentieth century, the govern-
progressively uncovers the critical bed. At first, the bed mental agencies in the UK that fund major coastal de-
daylights high in the slope face, and the only possible land- fense and protection schemes decided to reconsider their
slide is a small one. As the slope retreats, the landslide policies regarding coastal defense and protection (see
increases in size and volume. Eventually, at extreme stages Clarke and Lee 2003). In part, this was driven by finan-
in erosion, the critical bed disappears below the toe of cial motives, but in part, it has been for scientific rea-
the slope and landsliding related to it can no longer oc- sons. Successful defense of soft cliffs from erosion in
cur. The progression when coastal retreat occurs through some places has depleted coastal sediments, and changed
a sequence which contains a slide-prone horizon (usually the littoral hydraulics to the extent that neighbouring
a plastic clay) dipping landward, follows the slide as it areas have suffered accelerated erosion. When taken in
enlarges while cliff retreat takes place. Eventually, the conjunction with recognition of the likely effects of
slide-prone horizon plunges beneath the foot of the sea (a) rising sea levels, (b) climate change, and (c) ongoing
cliff, and active sliding ceases. eustatic changes in land levels in SE England, the eco-
In contrast to the situation where the beds are hori- nomic and technical arguments have swung against fur-
zontal, and where parallel retreat of the slope is possible, ther new construction, or even maintaining existing
where the beds dip into the slope, the nature and morphol- schemes. Finally, the scientific and environmental lob-
ogy of the slides relate to the relative erosion of the slope. bies have caused some coastal areas to be designated and
It is possible to view this series of sections not merely listed as of such scientific importance that the presump-
as a progression in time at a particular location, but also tion is that they will not be defended by major engineer-
for them to be a series of sections along a valley side or ing works. Such designations include being listed as be-
coastline. Interestingly, the resulting slide would have its ing a Site of Special Scientific Importance (SSSI), and
least width where the slip surface daylights at a high level Area of Outstanding Natural Beauty (AONB), or being
in the slide, broadening as the bed containing the basal protected as a wildlife habitat under the European Wild-
shear progressively descends towards the foot of the slope, life Habitats Directive.
and very rapidly narrowing again as the critical bed de- Such conflicting interests, especially at a time of chang-
scends further, to a position below the toe of the slope ing climate and rising sea levels, guarantee that the man-
which does not permit sliding to occur. The plan shape agement of unstable slopes will never degenerate to a
of the landslide is therefore controlled by bedding in a simple problem in elementary geotechnics, but will re-
related way to the cross-sectional shape. main a challenge for the foreseeable future.
Chapter 2 An Overview of Landslide Problems in the British Isles, with Reference to Geology, Geography and Conservation 25

Gregory CH (1844) On railway cuttings and embankments; with

References an account of some slips in the London Clay, on the line of the
London and Croydon Railway. Minutes Proc Inst Civ Engrs 3:
Barton ME (1977) Landsliding along bedding planes. Bull Int Assoc 135145
Eng Geol 16:57 Hutchinson JN (1969) A reconsideration of the coastal landslides at
Barton ME (1984) The preferred path of landslide shear surfaces in Folkestone Warren, Kent. Gotechnique 19:638
over-consolidated clays and soft rocks. In: Proceedings 4th Inter- Hutchinson JN (1970) A coastal mudflow on the London clay cliffs at
national Symposium on Landslides, Toronto, 3, pp 7579 Beltinge, North Kent. Gotechnique 20(4):412438
Bromhead EN (1978) Large landslides in London Clay at Herne Bay, Hutchinson JN (1983) Methods of locating slip surfaces in landslides.
Kent. Q J Eng Geol 11:291304 Bull Assoc Eng Geol XX(3):235252
Bromhead EN, Ibsen M-L (2004a) Bedding-controlled coastal land- Hutchinson JN, Bhandari, RK (1971) Undrained loading: a funda-
slides in Southeast Britain between Axmouth and the Thames mental mechanism of mudflows and other mass movements.
Estuary. Landslides 1(2):131141 Gotechnique 21(4):353358
Bromhead EN, Martin PL (2004b) Three-dimensional limit equilib- Hutchinson JN, Bromhead EN (2002) Keynote paper: Isle of Wight
rium analysis of the Taren landslide. In: Jardine RJ, Potts DM, landslides. International Conference on Landslides and Coastal
Higgins KG (eds) Advances in geotechnical engineering: the Management. Thomas Telford, London, 1, pp 370
Skempton Conference. Thomas Telford, London, 2, pp 789802 Hutchinson JN, Gostelow TP (1976) The development of an aban-
Bromhead EN, Chandler MP, Hutchinson JN (1991) The recent his- doned cliff in London Clay at Hadleigh, Essex. Philos Trans R
tory and geotechnics of landslides at Gore Cliff, Isle of Wight. Soc London, Ser A 283:557604
Slope Stability Engineering Developments and Applications, Hutchinson JN, Bromhead EN, Lupini JF (1980) Additional ob-
Thomas Telford, pp 189196 servations on the Folkestone Warren landslides. Q J Eng Geol
Bromhead EN, Hopper AC, Ibsen M-L (1998) Landslides in the Lower 13:131
Greensand Escarpment of South Kent. Bull Eng Geol Environ Hutchinson JN, Chandler MP, Bromhead EN (1981) Cliff recession
57(2):131144 on the Isle of Wight S-W coast. In: Proceedings 10th International
Bromhead EN, Ibsen M-L, Papanastassiou X, Zemichael AA (2002) Conference on Soil Mechanics and Foundation Engineering,
Three-dimensional slope stability analysis of a coastal landslide Stockholm 3(12):429434
at Hanover Point, Isle of Wight. Q J Eng Geol Hydroge 35:7988 Hutchinson JN, Bromhead EN, Chandler MP (1991) Investigations
Brunsden D, Jones DKC (1976) The evolution of landslide slopes in of the landslides at St Catherines Point, Isle of Wight. In: Chan-
Dorset. Philos Trans R Soc London, Ser A 283:605631 dler RJ (ed) Slope stability engineering developments and ap-
Burland JB, Longworth TI, Moore JFA (1977) A study of ground plications. Thomas Telford, London, pp 169179
movement and progressive failure caused by a deep excavation Ibsen M-L, Bromhead EN (1999) Head scarps and toe heaves. Z
in Oxford Clay. Gotechnique 27:557591 Geomorphol, special issue
Chandler MP, Hutchinson JN (1984) Assessment of relative slide haz- Jones, DKC, Lee EM (1994) Landsliding in Great Britain. HMSO, Lon-
ard within a large pre-existing coastal landslide at Ventnor, Isle don
of Wight. In: Proceedings 4th International Symposium Land- Kelly JMH, Martin PL (1985) Construction works on or near Land-
slides, Toronto, pp 517522 slides. In: Morgan CS (ed) Landslides in the South Wales Coalfield.
Clark AR, Fort DS, Davis GM (2000) The strategy, management and Polytechnic of Wales, pp 85106
investigation of coastal landslides at Lyme Regis, Dorset, UK. In: Lee EM, Jones DKC (2004) Landslide Risk Assessment. Thomas
Bromhead E, et al. (eds) Landslides in research, theory and prac- Telford, London
tice, Vol. 1, Thomas Telford. London, pp 279286 Margottini FA, Vai GB (2004) Litho-palaeoenvironmental maps of
Collin A (1856) (trans. W. Schriever) Landslides in clay. University Italy during the last two climatic extremes. Museo Geologico Gio-
of Toronto Press vanni Capellini, Bologna
Cooper MR, Bromhead EN, Petley DJ, Grant DI (1998) The Selborne Petterson KE (1955) The early history of circular sliding surfaces.
cutting slope stability experiment. Gotechnique 48(1):83101 Gotechnique 5:275296
Cruden DM, Thomson S, Hoffmann BA (1991) Observation of gra- Pitts J (1983) The temporal and spatial development of landslides in
ben geometry in landslides. Slope Stability Engineering Devel- the Axmouth-Lyme Regis undercliffs, National Nature Reserve,
opments and Applications, Thomas Telford, pp 3335 Devon. Earth Surf Proc Land 8:589603
Dixon N, Bromhead EN (1991). The mechanics of first-time slides in Pitts J, Brunsden D (1987) A reconsideration of the Bindon landslide
the London Clay Cliffs at the Isle of Sheppey, England. In: Chan- of 1839. Proc Geol Assoc 98:118
dler RJ (ed) Slope stability engineering developments and ap- Siddle HJ, Bromhead EN, Bassett MG (2000) Landslides and land-
plications. Thomas Telford, pp 277282 slide management in South Wales. Geological Series No 18. Na-
Dixon N, Bromhead EN (2002) Landsliding in London Clay Coastal tional Museum of Wales, Cardiff, Wales
Cliffs. Q J Eng Geol Hydroge (35)4:327343 Terzaghi K, Peck RB (1948) Soil mechanics in engineering practice.
DOE/GSL (Department of the Environment, UK/Geomorphological John Wiley & Son, New York
Services Ltd.) (1987) Review of research into landsliding in Great Trenter NA, and Warren CD (1996) Further investigations at the
Britain (12 vols) Folkestone Warren landslide. Gotechnique 46(4):589620
Chapter 3

Considerations about the Mechanics of

Slow Active Landslides in Clay

Luciano Picarelli

Abstract. Slow active landslides in clay include slides, mudslides and Movement is a result displacement along internal discon-
spreads. Movement is induced by any change of effective stress and tinuities (as the slip surface) and of internal strains. The
by creep; in the very long-time, some role may be played by a change
landslide body is generally subjected to a constant driv-
in soil properties. Looking at geological phenomena causing move-
ment, pore pressure fluctuations and erosion have a strong influ- ing force due to self-weight, thus movement is triggered by
ence on shallow translational slides and mudslides, while creep or change of the resisting force caused by change of boundary
erosion and other geological phenomena of stress relief govern conditions, and by viscous deformations. Furthermore,
movement of deep seated slides and spreads. In several cases, ex- in the long-term, any change of soil properties due to
cess pore pressures generated by changes of boundary conditions
weathering or to other processes of soil deterioration, can
may play a significant role.
play some role. Finally, change of slope morphology caused
Keywords. Clay, slide, mudslide, spread, active slow slope movement by movement itself may restrain further movement.
Therefore, the displacement rate depends on geometric
features of the landslide body, on the rate of effective stress
change and of soil deterioration and on viscous proper-
3.1 Introduction ties of both landslide body and discontinuities.
Usually, the expression active landslide refers to
Slow landslides are widespread in geomorphological con- (those) soil bodies which have experienced a general fail-
texts where stiff clays or clay shales crop out. According ure (in the sense discussed by Urciuoli et al. 2007) and
to old chronicles it can be argued that there are landslides are still moving. Their behavior is represented by the last
which have been active from thousands of years. part of the schematic diagram reported in Fig. 3.1, which
The risk posed by slow slope movements is rather low, depicts a typical time-slope displacement relationship
but management of landslides interacting with urban ar- from the so-called pre-failure stage, i.e. before general
eas, infrastructures and lifelines raises peculiar problems slope failure, to the post-failure stage (Leroueil et al. 1996).
since, even though the evacuation of people is not a press- Slow slope movements have much a longer life than rapid
ing need, in the long-term movements can severely dam- landslides: in some cases they last centuries or even thou-
age structures or interrupt the serviceability of lifelines, sands of years. Even though movement can appear per-
while slope stabilization or reinforcement of exposed el- manent, i.e. characterised by a constant velocity, very of-
ements is too expensive or non-effective. As a consequence, ten it displays a continuous change of the displacement
land management requires a complex and delicate cost-ben- rate, as a function of pore pressure fluctuations.
efit approach in which assessment of future slope behavior According to the classification proposed by Cruden and
and comparison among the effects of different remedial Varnes (1996), landslides can be classified slow as far as
measures take a crucial role. their velocity is less than 13 m month1, but for a velocity
This paper concerns natural slopes in clay, and is mainly less than 16 mm yr1, they are categorised as extremely
based on the state-of-the art report presented by L. Picarelli slow. However, here, the adjective slow will be used in a
and C. Russo at the 9th International Symposium on Land- flexible sense, to indicate a long-lasting movement un-
slides (Rio de Janeiro 2004), but includes further data and able to provoke, in the short-time, any significant conse-
considerations drawn from more recent papers and reports. quence on people and goods.
Many active landslides in clay (slides and mudslides)
present a slip surface, generated by shear strain localiza-
3.2 General Features of Slow Active Landslides tion and soil rupture, over which the soil mass slides ad-
vancing on the slope. However, just after failure, mudslides
Slow active landslides involve a number of natural slopes display a peculiar flow-like style attaining a rapid to mod-
much larger than what would be expected. The main types erate velocity followed by a slow decline, while the style
of slow active landslides are slides, mudslides and spreads. of movement progressively turns to slide (Picarelli 2001).
28 Luciano Picarelli

Fig. 3.1.
Simplified scheme of slope
movement from pre-failure to
arrest (after Picarelli 2000)

In some cases people include in the category of land-

slides soil masses which did not experience any general
slope failure, but are subjected to continuous movement
due to perceivable internal strains. For instance, lateral
spreads caused by valley rebound can be driven by defor-
mation of even large soil masses, but not by slope failure.
This implies that a persistent slip failure does not neces-
sarily exist. Also movements caused by creep or by inter-
nal soil deformation provoked, for instance, by ongoing
progressive failure or by internal stress redistribution,
reveal a pre-failure more than a the post-failure stage.
Moreover, these phenomena will not necessarily lead to Fig. 3.2. The nature of movement of slides (from Picarelli and Russo 2004)
general slope failure.
tively large. In other cases formation of the shear zone is
a more complicated process because of non uniformity
3.3 Considerations about the Mechanics of the state of stress. Generally it can be said that it starts
of Active Slides and Mudslides forming in the most stressed part of the slope, propagat-
ing into the soil mass as a result of a mechanism of pro-
3.3.1 Active Slides gressive failure which culminates in the general failure
(Urciuoli and Picarelli 2004).
Slow active slides generally move over a shear zone lo- In the post-failure stage, the mobilized soil mass can
cated at the base of the landslide body. This zone is the eventually move over the slip surface. Even if immedi-
result of a complex process of strain localisation occur- ately after failure the landslide may experience a strong
ring in the pre-failure stage, whose final effect is the for- acceleration because of brittle soil behavior, eventually it
mation of a shear surface, also called slip or sliding sur- progressively slows down due to change of both slope
face, internal to the shear zone. morphology and Generalised Brittleness Index of soil
Urciuoli (2002) analyses the formation of the shear (DElia et al. 1998), while the shear strength along the slip
zone in the case of infinite slope, showing that is accom- surface approaches the residual value. At residual, the re-
panied by rotation of the principal stresses and forma- sisting force along the slip surface depends only on the
tion of minor shears (Skempton 1967; Morgenstern and normal effective stress and, possibly, on viscous effects
Tchalenko 1967). As the direction of principal stresses associated with the displacement rate. In the final stage,
becomes consistent with the direction of the theoretical i.e. once the friction angle has attained a constant value,
failure plane, i.e. parallel to the ground surface, a persis- slope movement is governed by the cyclic balance between
tent slip surface eventually forms and slope failure takes driving force and resisting force, being the result of slid-
place. This occurs simultaneously in all points of the fail- ing along the slip surface and of internal strains (Fig. 3.2).
ure plane because of uniformity of the state of stress. The Slipping occurs every time the residual strength is mobi-
thickness of the shear zone depends on the initial state of lized along the shear surface, for instance when pore pres-
stress and on the shear strength of soil: for a critical value sure trespasses a critical value. Internal strains are caused
of the coefficient of earth pressure at rest, it is nil; for high by changes of the effective stress field and by viscous phe-
values of OCR it is quite small; for small values, it is rela- nomena, but in the very long-term additional strains can
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 29

be provoked by change of stiffness due to soil deteriora- ing balance between the driving and resisting force due
tion. Concerning this point, Picarelli (2000) shows that to modifications of the morphology caused by movement
the residual strength of clay shales may decrease with time itself (DElia et al. 1998). As a consequence, circular slides
because of changes of grain size due to progressive break- generally run short distances and are active for quite a
ing of bonds linking aggregates of clay particles. In addi- short time.
tion, accounting for data provided by Di Maio (1996) and In clay deposits the groundwater table is located at
by Di Maio and Onorati (2000), Picarelli et al. (2006) il- rather a small depth from the ground surface, and fluctu-
lustrate possible long-term effects of infiltration of fresh ates during the year as a consequence of seasonal water
water in marine clay of high plasticity leading to a de- recharge and discharge. In uniform clay, the magnitude
crease of both peak (softening) and residual shear of pore water fluctuations decreases with depth because
strength, but certainly, even soil stiffness experiences of the time required by pore pressure to reach equilib-
some decay. Some numerical analyses on the effects of rium with boundary conditions, compared to the dura-
soil deterioration on slope movements are reported by tion of the wet season. Kenney and Lau (1984) present
Yoshida (1990) through a manipulation of the Hoek and excellent experimental data regarding a slope in Ontario,
Brown (1980) strength criterion. while Cavalera (1977) reports analytical solutions of this
The behavior of translational slides (an example is re- problem for sinusoidal changes with time of hydraulic
ported in Fig. 3.3a) is governed by the component of the conditions at the ground surface. This suggests that pore
driving force in the direction of movement. Since such a pressure fluctuation due to rainfall (or to ice melting)
component is essentially constant, the displacement rate strongly governs the behavior of shallow translational
depends on changes of the resisting force caused by slides, while its role is minor for deep-seated landslides.
changes of effective stress and on soil viscosity. Quite a In contrast, the behavior of deep seated slides should be
similar framework held for compound slides, most of essentially governed by creep, i.e. by soil deformation
which are characterised by an almost rectilinear and sub- occurring under an essentially constant state of stress, or
horizontal lowermost part of the slip surface (Fig. 3.3b). by stress changes due to geological phenomena, as ero-
However, in this last case the non uniformity of the state sion. A well known slide driven by fluvial erosion is the
of stress may determine a complex interaction between huge La Frasse slide, which has an extension of 42 km3.
the rear and the front part of the landslide body. The Monitoring dates back 170 years, since construction of the
behavior of circular slides strongly depends on chang- major cantonal road connecting the town of Aigle to the

Fig. 3.3. Examples of slow active slides. a The essentially translational Fosso San Martino slide (from Bertini et al. 1986); b the compound
Castle Hill slide (from Varley and Warren 1995)
30 Luciano Picarelli

Col des Mosses Pas, but the landslide is much older, as of years presented only some very small cracks, while only
dating of wood fragments suggests (Tacher et al. 2005). the oldest constructions (a monastery built some centu-
Literature reports numerous examples of slow slides. ries ago, a building and a high retaining wall) displayed
A number of these, as the La Frasse slide, are have been large cracks. Monitoring through some inclinometer pur-
active for hundreds or even thousands of years, moving posely installed in the area revealed the presence of a slow
with an average displacement rate as low as a few centi- active slide whose slip surface is located just at the top of
meters per year or less. An example of long-lasting land- clay shales. The displacement rate was quite constant
slide is Castle Hill slide shown in Fig. 3.3b, which is be- (around 1 cm yr1) and uniform all over the area (Fig. 3.5).
lieved to have formed about 10 000 years ago. Its displace- Readings carried out with Casagrande piezometers indi-
ment rate in the last tens of years, argued from deforma- cated that the groundwater table is located at the base of
tion of a pipeline located within the landslide body, is the landslide body, some decimeters above the top of clay,
about 12 mm yr1. The displacement rate of long-last- and that it experiences only very little annual fluctuations
ing slow movements is often assumed to be uniform with because of high permeability of the debris, preventing
time, mainly because of a lack in continuous measure- formation of an aquifer. Therefore the state of stress is
ments, but once more data are available, some landslides practically constant and no significant future changes of
show themselves to be intermittent (stick slip move- this can be predicted. The small internal deformation of
ments) being driven by pore pressures changes occur- the landslide body depends on the high stiffness of de-
ring over very short periods of time (Van Genuchten bris; moreover, the absence of significant differential
1984). This naturally disproves the assumption of creep movements can explain the absence of severe damages
as cause of movement. on dwellings and walls built recently, while large cracks
An example of creeping slide is reported by Picarelli across the walls of the monastery and of the other old
and Simonelli (1991), who discuss deformation phenom- structures testify deformations cumulated over a much
ena affecting the neighborhood of a small Italian town longer period of time.
(Fig. 3.4), located on a gentle slope. The buildings rise on Another example of slope movement presumably caused
coarse calcareous debris covering a deposit of highly fis- by creep is reported by Corominas et al. (2005), who de-
sured sheared clay shales. At the time of investigations, scribe the Vallcebre landslide in fine grained stiff deposits
the majority of dwellings and walls built in the last tens (Fig. 3.6a,b). Figure 3.7 reports the water levels measured

Fig. 3.4. A urban area experiencing slow movements and displacement vectors between 1987 and 1990 (after Picarelli and Simonelli 1991)
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 31

Fig. 3.5.
Cumulated displacements in
the urban area of Fig. 3.4 (from
Picarelli and Simonelli 1991)

Fig. 3.6.
The Vallcebre slide (from Coro-
minas et al. 2005): (a) plan,
(b) longitudinal sections

with a standpipe piezometer and displacements at vertical 1997, the displacement rate was around 15 mm month1,
S2 (Fig. 3.6a). It shows that in some periods of the year pore pressure being constant, but a sudden increase of
pore pressures remain constant, even though the land- this, followed by a decrease to a value slightly higher than
slide moves anyway. For instance, between April and June, before, determined a doubling of velocity. The Authors
32 Luciano Picarelli

Fig. 3.7. Groundwater level, rainfall and displacements at vertical S2, Vallcebre landslide (from Corominas et al. 2005)

Fig. 3.8.
Fosso San Martino slide.
a Ground-water levels; b dis-
placements of the ground sur-
face (from Bertini et al. 1986)

assume that the increase of the displacement rate is con- movement stops. In wetter periods, it re-starts following
trasted by increase of the residual shear strength due to pore pressure rising. As remarked by other Authors
rate effects. This idea is shared by other Authors, who re- (Cartier and Pouget 1988; Corominas et al. 2005), the dis-
port similar considerations about the behavior of soils placement rate increases as the pore pressure increases,
involved in slow movements (Vulliet 1986; Angeli et al. but the relationship between pore pressure and displace-
1996; Vulliet and Hutter 1998). ment rate is non-linear. In addition, Bertini et al. show
A more complex viscous soil behavior is hypothesised that, for the same value of pore pressure, the velocity dis-
by Bertini et al. (1986) for the Fosso San Martino slide, in played during groundwater rising is higher than during
fine grained colluvial soils (Fig. 3.3a). Figure 3.8 reports lowering. Hence, two different relationships can be es-
some results of monitoring covering a period of 6 years. tablished between pore pressure and displacement rate;
Movement is directly correlated to rainfall and to conse- for the same reason, reactivation and arrest occur for dif-
quent pore pressures changes, starting once the mobilized ferent pore pressure thresholds (Fig. 3.9a). The Authors
shear strength along the slip surface is about 95% of the attribute such a behavior to the viscous behavior of soil
residual value. When pore pressure decreases below a along the slip surface, accounting for both primary and
threshold value, as between summer, 1981, and fall, 1983, secondary creep. In fact, they argue that:
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 33

Fig. 3.9. Relationship between displacement rate and water level (from
Bertini et al. 1986)

1. during the stage of pore pressure rising, the displace-

ment rate is the result of two contrasting phenomena:
increase of the rate of primary creep, which is associ-
ated with increasing stress level, and decrease of the
rate of steady-state creep, which depends on time;
2. in the stage of pore pressure decrease, the two effects
have the same negative sign and the resulting velocity
is smaller.

The existence of different thresholds for increasing and

decreasing pore pressure is stressed by other Authors too,
but sometimes their observations are opposite, suggest-
ing different interpretations. Moore and Brunsden (1996),
as Bertini et al., find that the threshold pore pressure re-
quired to reactivate the Worbarrow Bay mudslide, Dorset,
is lower than the one required to stop movement, but they
explain this apparent inconsistency with a change in pore
water chemistry during periods of rest. They also note
that each reactivation requires a pore pressure larger than
previous reactivations. Differently, in the case of the Alver
mudslide, Angeli et al. (1996) note the existence of two
different thresholds corresponding to reactivation and Fig. 3.10. Displacements (a) of the Miscano mudslide (from Picarelli
arrest, but in this case the lowest one corresponds to ar- et al. 1999), and relationship (b) between velocity divided by ground-
rest. They assume that the higher threshold required for water level and groundwater level (from Mandolini and Urciuoli 1999)
reactivation is due to some increase of the residual
strength occurring during the period of rest. seems to occur once the groundwater table reaches the
A non linear relationship between pore pressure and ground surface, i.e. for a mobilized friction angle of 18.
displacement rate has been found by Mandolini and The ratio between the first stress level (creep threshold)
Urciuoli too (1999) for the Miscano mudslide in softened and the second one (full shear strength mobilization) is
clay shales (Fig. 3.10). According to Fig. 3.10b, movement then only 70%, which seems to be very low when com-
starts when the groundwater table is located at a depth pared to data provided by Bertini et al. (95%) and to labo-
around 2 m: in this condition, the mobilized friction angle ratory experiences (Boucek and Pardo-Praga 1984). This
is about 13. A full shear strength mobilization, corre- raises some doubts about the real mechanisms of move-
sponding to a high displacement rate (tending to infinite), ment that will be discussed below.
34 Luciano Picarelli

Every assumption has a conceptual model behind. should be checked by laboratory tests. Unfortunately,
Referring to slopes, many researchers have the rigid-plas- available data do not help very much. Through tests on
tic constitutive law in their mind. According to this model, clay, Kenney (1967) and Skempton (1985) remark a neg-
the landslide body should move along the slip surface as ligible or little increase of the residual shear strength with
a block, with a velocity depending on the constitutive law the displacement rate. Furthermore, no influence is no-
of the slip surface. ticed by Hungr and Morgenstern (1983) for sand, while
Assuming that the residual friction angle is the one data collected by Picarelli and Urciuoli (1988) and by Tika
mobilized just at the beginning of movement (13 in the et al. (1996) through laboratory tests on clay, are more
case of the Miscano mudslide), a conceptual model ca- problematic, suggesting that in some cases excess pore
pable to justify the landslide behavior resembles the one pressure (either negative or positive) can be induced by
proposed by Corominas et al., which assumes a rate-de- fast movement. This adds a complication in the model,
pending residual strength along the slip surface. There- which should account for drainage conditions. Further-
fore, once a unbalanced force (difference between the driv- more, in this case viscosity is not necessary to explain
ing and the resisting force due to basal friction) estab- the slope behavior, because increasing shear strength
lishes, the soil mass tends to accelerate, but this is con- would essentially be a consequence of induced negative
trasted by growing resistance along the slip surface: the pore pressure.
net effect depends on both the magnitude of the unbal- If we assume that the true residual friction angle is the
anced force and the rate of shear strength increase. Natu- one which is mobilized when the displacement rate at-
rally, the dependence of the residual strength on the dis- tains an infinite value (18 in the case of Miscano mud-
placement rate, thus the validity of proposed model, slide, when the piezometric level reaches the ground sur-
face), the model to be adopted for interpretation of the
landslide behavior should include pre-failure creep (as
assumed by Bertini et al.), which is able to justify move-
ments occurring for pore pressures less than the critical
value required to provoke a full shear strength mobiliza-
tion. Therefore, once again, viscosity of the slip surface
could play a fundamental role, but in this case, prior to
mobilization of the shear strength, which does not neces-
sarily depend on the rate of movement. In this last case, a
full shear strength mobilization would lead to catastrophic
The two models of slope behavior could be easily uni-
fied, but both consider the soil mass as a block moving
along a viscous interface. Reality is more complex since
soil is deformable and both initial and induced state of
stress are not uniform in the slope: as a consequence, in-
ternal strains may play a significant role on landslide be-
havior. This last point is stressed by Picarelli et al. (2004)
who argue that in many cases the landslide body is not
entirely mobilized by pore pressure fluctuations, which
cause internal strains associated with slipping along only
part of the sliding surface.
Figure 3.10a shows that the displacement field of the
Miscano mudslide at a given time is not uniform along
the slope, with decreasing values downward. This can be
explained accounting for pore pressure changes in the wet
season (Fig. 3.11a), which are faster upslope than downs-
lope, causing a temporary mobilization of the uppermost
part of the landslide only, while the local shear stress
mobilized at the base of the lowermost remains lower than
the shear strength. This causes a compression of the land-
slide body induced by the overstress caused by restraint
Fig. 3.11. Analysis of displacements of the Miscano mudslide.
imposed by the lowermost part of it. Such a phenomenon
a Adopted model; b measured groundwater levels; c measured and has been simulated by the FEM. The slip surface has been
calculated displacements (from Picarelli et al. 1999) characterised with elastic-perfectly plastic elements along
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 35

which is operative the residual shear strength, while the ditions or non uniform geometric and mechanical param-
landslide body has been modeled with an elastic-plastic eters of soil, have been investigated by Russo (1997) and
constitutive law (Cam-Clay) adopting the parameters re- discussed by Picarelli and Russo (2004) using both an elas-
ported in Fig. 3.11b. Slope displacements have been cal- tic-plastic and an elastic-viscous-plastic constitutive law
culated for the same pore pressure increase which has been of soil. In particular, the use of an elastic-viscous-plastic
measured in site, as in Fig. 3.11a. The results of the analysis constitutive law allows to perform more sophisticated
are comparable to reality (Fig. 3.11b): in particular, sig- analyses. As an example, a slope model as the one used to
nificant movement develop even though the safety factor interpret movement of the active Miscano mudslide, can
of the lowermost part of the mudslide body remains higher account for the effects of relaxation occurring in any phase
than one. Therefore, this simple model confirms that of pore pressures decrease (dry season), when the com-
movement can be essentially caused by internal strains pressive state of stress stored in the landslide body in pre-
more than by full mobilization of the landslide body. vious wet season (when pore pressure increases) progres-
The same mechanism can also explain the non linear sively disappears as a consequence of stress relaxation. It
increase of displacement with increasing pore pressure, is worth noting that the magnitude of deformation oc-
discussed above (Figs. 3.9a and 3.10b): in fact, this result curring in the wet season depends on viscous soil prop-
depends on the increase of the length of the part of the erties and on time span prior to the new pore pressure
landslide body which is mobilized by pore pressure increase. Such a deformation history governs the magni-
changes (Picarelli and Russo 2004). In addition, since pore tude of pore pressure (threshold) required to activate new
pressures vary with time, movement too varies with time, movements, which decreases as the magnitude of relax-
so that soil viscosity is not necessary to explain the rate- ation increases.
depending behavior of slide. Similar considerations could be made about opposite
Naturally, previous simplified interpretation of slope mechanisms inducing extensive strains in the slope.
behavior does not exclude mobilization of a viscous com- All these remarks about the possible role of soil
ponent of displacement within the landslide body and deformability on slope movement appear consistent with
along the slip surface, which can be easily incorporated data provided by some Authors (Wilson 1969; Jappelli et al.
in the model (Vulliet 1986). 1977; Nakamura 1984; Pouget 1996).
Similar effects on slope behavior, i.e. displacement
caused by internal deformation of the landslide body, can
be provoked by uniform pore pressure increase, if move- 3.3.2 Active Mudslides
ment is locally constrained by:
The typical flow-like style of mudslides is revealed just at
 variation of the geometry of the landslide body, be- failure or immediately after failure, when the landslide
cause of increasing thickness and/or decreasing slope body displays a high mobility, spreading along the slope
of the slip surface moving from upward to downward; and filling pre-existing tracks. In this stage the mobilized
 local change of the width of the landslide (3D effect), soil mass moves quite rapidly over the ground surface,
as in mudslides which present a neck in between the eroding and incorporating the top soil and the upper part
alimentation zone and the main track; of the underlying deposit: as a consequence, it can expe-
 any change of the shear strength parameters along the rience some subsidence (Hutchinson 1970; Corominas
slip surface. 1995). The described mechanism of movement implies
that the base of the mudslide body is subjected to high
These considerations offer alternative scenarios about shear stresses which are presumably responsible for in-
slope behavior. In fact, at least in the case of long land- tense remoulding of quite a thick basal slice of soil. In
slides, even very small strains, when integrated to the en- stages following first post-failure movements, the soil
tire length of the landslide, could give rise to a significant mass decelerates assuming a slide style, but the occur-
component of displacement. As an extreme case, for a rence of previous flow-like mechanisms can be still eas-
100 m long landslide having a free upper boundary (the ily recognized from the assumed morphology of the land-
crown) and a fixed lower boundary (the toe), an average slide body. This typically displays an alimentation zone,
compressive strain of 102% due to pore pressure rising, a track and a fan-shaped accumulation zone. In this last
can trigger a displacement of the uppermost part of the stage of slope deformation prior to definitive arrest, a well
landslide body of 1 cm, solely due to compression. It is defined persistent slip surface, which could not survive
worth noting that, in stiff clay, a strain equal to 102% can to movements occurring in the flow-like stage, presum-
be provoked by increase of pore pressure in the order of ably governs the mudslide behavior.
1 kPa, i.e. by a rising of the water table in the order of 10 cm. Mudslides are bounded by a main scarp that domi-
Described mechanisms of slope movement, which nates a depletion area (Fig. 3.12). This alimentation zone
might be governed by constraints due to boundary con- feeds the landslide body with fresh soil masses (surges),
36 Luciano Picarelli

Fig. 3.12. The Brindisi di Montagna mudslide (from Cotecchia et al. 1984)

mobilized by local failures of the scarp, which are con-

veyed into the main track, causing periodical flow-like
reactivations of movement. Further alimentation is pro-
vided by slips occurring along the flanks of the track.
Surges form secondary accumulations along the track, but
can reach the toe of the slope, thickening the accumula-
tion zone. The presence of independent active or tempo-
rarily quiescent landslide bodies within the area occupied
by the mudslide body is revealed by inclinometer profiles
which show more shear zones along the same vertical.
To sum up, the main features of mudslides concern:

 their morphology, which reveals the mechanism of

movement which characterises the initial post-failure
flow-like phase: when channellized within a pre-exist-
ing track, long-term movements are restrained and Fig. 3.13. Typical displacement profiles of a mudslide. a In the first
governed by a complex 3D condition; stage of movement (flow-like style); b in the last stage of movement
 the features of the soil mass, which is highly remoulded (slide style) (from Comegna et al. 2005)
as a consequence of high deformations and softening
experienced during first movements (Hutchinson 1988; Dissipation of excess pore pressures leads to a de-
Picarelli 1993); crease of the rate of movement while the soil mass takes
 the thickness and fabric of the shear zone: while in the the features of a slide: in this stage movement localizes
case of slides this is quite thin (generally a few centi- along a slip surface, while shear strains within the land-
meters) and displays a set of minor shears, in the case slide body become smaller than in previous flow-like
of mudslides it is rather thick (up to one meter) and stage (Fig. 3.13). However, some data suggest that even
fully remoulded (Picarelli et al. 2005): often the struc- this sliding phase can be characterised by development
ture of the parent formation is completely obliterated. of moderate excess pore pressures because of internal
stress changes.
Several Authors (Hutchinson and Bandhari 1971; Figure 3.14 shows pore pressures measured within
Picarelli 1988;) argue that the initial flow-like style is a the Masseria Marino mudslide at a depth of about 3 m.
consequence of building-up of positive excess pore pres- Figure 3.13a concerns a quiescent zone and Fig. 3.13b an
sures. Pellegrino et al. (2004) discuss the main mecha- active zone subjected to slow movement. While in the
nisms which are supposed to be responsible for trigger- quiescent zone pore pressures display smooth seasonal
ing of excess pore pressures. They include: fluctuations, quite in a good agreement with the results
of numerical simulations carried out using pluviometer
 rupture itself; data and adopting the permeability of soil obtained by
 seismic loading; in situ and laboratory tests, in the most active zone pore
 accumulation of debris over a pre-existing mudslide pressures display rapid fluctuations which do not seem
body as a consequence of secondary failures or of rapid to follow a logic course (Comegna et al. 2004). The most
erosion along the main scarp or flanks of the track; reliable explanation is that any local soil deformation can
 loading caused by surges traveling over the mudslide trigger excess pore pressure: where deformation is by
body. compression, induced pore pressure is positive.
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 37

Fig. 3.14. Masseria Marino mudslide: location of the piezometers considered in the analysis (a, b); calculated and measured water levels in
quiescent (c, d, e) and active zones (f, g) of the mudslide (after Comegna et al. 2005)

Such a mechanism has been checked through a sim- parameters adopted in the analysis have been obtained
plified analysis. The mudslide body has been modeled by the best fitting of laboratory tests (Comegna 2005):
with a non linear elastic constitutive law characterized their values are reported in Table 3.1.
by an isotropic yielding law (Soft-Soil Model present in The sliding surface located at the base of the shear zone,
the library of the code PLAXIS). In addition, has been has been simulated by interface elements along which is
considered a basal 1 m thick softer shear zone. The soil operative a residual friction angle of 13. The analysis
38 Luciano Picarelli

Fig. 3.15.
Plastic active zone formed as a
consequence of pore pressure
rising (from Comegna et al. 2007)

Fig. 3.16. Displacements, longitudinal strains and excess pore pressures at three sections of the mudslide (from Comegna et al. 2007)
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 39

simulates the pore pressure rising induced by rainfall, by undrained compression. However, the longitudinal strain
imposing a water film along the ground surface (pore pres- L is not uniform because of the different stiffness of
sure equal to zero). During this phase, only part of the mudslide body and shear zone; this difference in stiffness
slip surface is mobilized, but the length of the mobilized is also responsible for lower excess pore pressure in the
part propagates downward as pore pressure increases. As shear zone. Longitudinal strains, pore pressures and dis-
a result, the active mudslide body slides downward, placements at three different sections of the slope are re-
compressing the still stable part located ahead; at the same ported in Fig. 3.16.
time, an active zone forms immediately upslope (Fig. 3.15). During the following stage of analysis, excess pore pres-
In order to simulate local soil rupture in the active zone, a sures are allowed to equalize. This phase is characterized
vertical cut has then been imposed. Assuming that the by the overlap of two opposite phenomena: dissipation
deformation of the landslide body after cracking is fast of excess pore pressure and continuing rising of water
enough to trigger excess pore pressures, the concerned table due to infiltration. Figure 3.17 reports the evolution
stage of analysis has been performed assuming short- of pore pressure at two points, one located in the shear
term (undrained) conditions. As a consequence of crack- zone and the other one in the mudslide body, in three sec-
ing, the mobilized part of the sliding surface further tions of the landslide. Because of the high gradient of the
propagates downward while pore pressures rise due to piezometer head closely around the two points, excess
pore pressures rapidly dissipate: a decrease in the mud-
slide body is associated with increase in the shear zone.
Such a result could explain the anomalous drops of pore
pressures monitored in the Masseria Marino mudslide
(Fig. 3.14b).

3.4 Consideration about the Mechanics

of Active Lateral Spreads

Lateral spreads represent a special category of slow active

landslides in clay. They are so slow to be recognized only
through sophisticated monitoring or from continuing
damage to old structures. According to literature, lateral
spreads can be triggered by tectonic uplift, by removal of
lateral support following glacial retreat or by river ero-
sion (Radbruch-Hall et al. 1976).
An interesting example is movement which involves
the Monte Verna hill (Fig. 3.18) where San Francis retired
in the last years of his life (Canuti et al. 1990). The hill is
constituted by jointed massive calcarenites and bedded
sandstones which rest on a deposit of highly fissured
sheared clay shales. It is subjected to lateral movements,
essentially driven by squeezing out of clay shales. Move-
ment causes opening of vertical joints in the rock slab.
A schematic description of the mechanisms which
govern spreading induced by valley formation is shown
in Fig. 3.19. This concerns the case of a rock slab resting
on clay, as in Fig. 3.18. Erosion starts developing along
fractures of the slab (stages 1 and 2). The stress relief
caused by unloading provokes rising of the valley floor;
as a consequence, the clay deforms forming an anticline
(stages 3 and 4). Further deepening of the valley reaches
the buried top of clay, which squeezes out bringing about
lateral deformation of slopes. As a consequence of im-
posed drag forces, a shear zone forms at the contact be-
tween clay and rock bed, and the slab is subjected to frac-
Fig. 3.17. Pore pressure evolution at two points within the mudslide turing as a consequence of tensile stresses imposed by
body (from Comegna et al. 2007) shear (stage 5). Induced movement causes opening of
40 Luciano Picarelli

Fig. 3.18. Deformation phenomena of the Monte Verna mountain (from Canuti et al. 1990).

vertical joints which can be filled by clay spread from the

bottom. In addition, internal rock blocks subside, while
lateral blocks tilt, forming a trench at the top of the slab.
Further movements trigger landslides along the bound-
aries of the slab contributing to their complete split up
(stages 6 and 7).
The geological phenomena which govern such a pro-
cess, even when very fast in geological terms, are extremely
slow in the human perception, thus consequent move-
ments are extremely slow. However, in case of very thick
clay deposits, unloading due to erosion can be slower than
the time required for dissipation of induced negative ex-
cess pore pressures (Neuzil 1993), causing a delay of as-
sociated deformations. In seismically active areas, earth-
quakes can add their effects to erosion and creep, pro-
ducing acceleration of deformation.
Other cases of spreads can be found in the Italian sci-
entific literature (Crescenti et al. 1994; Fenelli et al. 1996;
Chelli et al. 2006). Several of them concern towns located
on the top of hills and mountains in the geologically ac-
tive area of Apennines. A well known example is the
Orvieto hill, in the Tiber valley (Lembo-Fazio et al. 1984;
Tommasi et al. 2006), which is constituted by a tuff slab
resting on stiff clay. A similar situation is described by
Tommasi and Rotonda (1995) who discuss the deforma-
tions of the calcareous San Leo cliff, on top of which stands
a Renaissances Castle where the famous Cagliostro was
hold in jail until his death.
An interesting case is the Bisaccia spread, in Southern
Italy (Di Nocera et al. 1995). Bisaccia rises on a slightly
cemented conglomerate slab with thickness exceeding
100 m, which rests on intensely fissured highly plastic
clay shales. In the last 300 000 years, the area has been
deeply eroded by downcutting along two parallel faults
(Fig. 3.20), through a geological process very similar to
the one described in Fig. 3.19. Concentrated erosion left
an elongated hill in the middle. Presently, the floor of
the two valleys is lower than the bed of the conglomerate
slab over which stands the town. The slab is divided into
large blocks separated by vertical cracks. Its boundaries
are subjected to landslides involving both upper con-
glomerates and lower clay shales. Some movements in-
Fig. 3.19. Effects of river erosion on the deformation pattern of a rock volve both conglomerate blocks and clay at the foot
slab resting on clay (from Pasek 1974). (slides); others involve only clay and generally display a
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 41

flow-like style (Picarelli et al. 2006). The top of the hill (1980), two verticals, one in the eastern valley and the
presents a series of morphological steps. In the past, these other in the urban area, were instrumented with
have been smoothed with man-made ground or bridged Casagrande and vibrating wire piezometers (Fenelli and
with stairs. In addition, buildings show fissures and Picarelli 1990; Di Nocera et al. 1995). Finally, on February,
cracks; old reparations reveal the occurring of a process 1981, a number of benchmarks were installed in the town
of general deformation. to monitor deformations induced by the quake. Monitor-
The area is shaken by earthquakes having a return time ing covered more than seven years, until October, 1988.
of some tens of years. In the last century, two strong earth- Figure 3.21 reports pore pressures measured along the
quakes occurred in 1930 and in 1980 (the last one charac- two instrumented verticals. Deficient values (locally nega-
terized by a Magnitude M = 6.3). Seismic events system- tive, i.e. below the atmospheric pressure) have been mea-
atically cause cracks in the masonry of old buildings and sured below the bed of the eastern valley where unload-
in pavements. The main effects of the 1930 and 1980 earth- ing has its major effect. In contrast, pore pressures above
quakes were a series of fractures on pavements. Most of the their theoretical value have been measured just below the
cracks generated by the two events coincide, and are located slab. However, accurate readings have shown a constant
along natural morphological steps; in addition, buildings decrease in time.
most severely damaged by the 1980 quake are located These results have been interpreted accounting for the
along such alignments (Fenelli 1986). This means that the geological phenomena affecting the site. In fact, account-
damages were mostly caused by opening of fractures. ing for the thickness and low permeability of clay shales,
The area has been systematically investigated since erosion is quite fast in geological terms and is assumed to
1981. The first campaigns included field and laboratory trigger negative excess pore pressures (Fenelli and Picarelli
investigations which allowed to obtain a complete me- 1990). Di Nocera et al. (1995) show that swelling of clay
chanical characterization of clay shales (Picarelli et al. can explain the geomorphological shape of the two val-
2002). In 1985 and 1989, thus after the Irpinia earthquake leys and the distribution of the water content in the sub-
soil; other consequences of erosion are cracking of the
slab, squeezing out of clays and a shear zone recognized
at the conglomerate-clay shale contact.
Formation of the hill has been simulated by FEM
analyses (Picarelli and Urciuoli 1993; Di Nocera et al.

Fig. 3.21. Pore pressures measured in the Bisaccia area (from Fenelli
Fig. 3.20. The Bisaccia hill (from Fenelli and Picarelli 1990) and Picarelli 1990)
42 Luciano Picarelli

1995) assuming that erosion started 300 000 years ago, velocity of one order of magnitude less. Horizontal dis-
starting from an elevation corresponding to the present placement at the contact between conglomerate and clay
top of the hill. Figure 3.22 shows calculated vertical dis- shale is shown in Fig. 3.23. Calculated movement is in
placements of the valley floor, at the depth of its present the order of 0.2 mm per century; the differential displace-
elevation, and of the top of the hill. The figure shows ment along the conglomerate-clay shale interface is re-
that the valley floor is rising with a velocity of about sponsible for formation of a shear zone. It is worth not-
0.04 mm yr1, while the hill sinks in clay shales with a ing that the deformations calculated in the analysis not
only depend on the assumed rate of erosion, but also on
the rate of pore pressure equalization.
Previous data about the effects of the 1980 earthquake
suggest that strong seismic events can accelerate movement
of the hill. Figure 3.24 shows the settlements measured from
February, 1981, to respectively February, 1982, and Octo-
ber, 1988, along a longitudinal section of the town (Fenelli
et al. 1992). The location of major and minor fractures in
the slab is represented by arrows. A profile of the hill along
the same section is also shown in a magnified scale.
The figure shows that:

 until October, 1988, the entire urban area underwent a

general subsidence with settlements increasing from
South to North, where the hill is truncated by a steep
high slope;
 settlements increased with time; eight years after the
quake, the maximum value measured along this sec-
tion was about 12 cm: the average settlement rate in
the examined period was about 1 cm yr1;
 the general subsidence of the hill is discontinuous,
probably because of independent movements of con-
glomerate blocks; in particular, some blocks experi-
ence a vertical translation, others a rotation;
 the trend of vertical displacements is very similar to
the morphological profile of the hill, that seems hence
determined by cumulated deformations induced by
erosion and by seismic events.

Despite the absence of further measurements after

1988 due to loss of benchmarks, the development of new

Fig. 3.22. Calculated vertical displacement of the valley floor (a) and Fig. 3.23. Calculated horizontal displacement at the conglomerate-
(b) of the top of the hill (Picarelli and Russo 2004) clay shale interface (Picarelli and Russo 2004)
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 43

Fig. 3.24.
Vertical displacements mea-
sured after the 1980 earthquake
and longitudinal profile of the
hill (from Fenelli et al. 1992)

cracks in buildings demonstrate that movements were

3.5 Conclusions
still continuing. Unfortunately, no data are available
about horizontal displacements, but certainly they were
not negligible. Typical slow active landslides in clay are translational
More information are provided by numerical analy- slides and mudslides, but other types of movements, as
ses (Olivares 1997; Lampitiello et al. 2001). The sim- spreads, also belong to the category of slow active landslides.
ulations have been carried out using seismogram Active slides and mudslides advance along a pre-ex-
data recorded in the town during the earthquake, de- isting slip surface and are driven either by changes of
convoluted to reproduce the seismic input motion at boundary conditions, as those caused by rainfall or ero-
the top of bedrock. The analyses demonstrate that, in sion, or by viscous deformations. The effects of pore pres-
a large part of the clay shale deposit under the slab, sure fluctuations prevail in the case of landslides of mod-
the quake induced shear strains larger than the volu- erate thickness, while those of erosion and of creep can
metric threshold, causing a local building up of posi- be significant in the case of deep-seated landslides. To
tive excess pore pressures. The dissipation of these is understand the mechanics of slides and mudslides, the
responsible for delayed vertical displacements, as shown model of rigid-plastic body is not adequate. In fact, since
above. any change of boundary condition provokes a non ho-
Summing up, the hill is subjected to two different geo- mogeneous variation of the stress field, internal defor-
logical phenomena: erosion and earthquakes. Both con- mation of the landslide body is a normal condition: since
tribute to slope movements. Erosion brings about a deep- slow landslides present very small displacements, inter-
ening of the valleys surrounding the hill. The superim- nal deformation is a significant component of movement.
posed effects of earthquakes cause a general subsidence However, available data show that positive excess pore
of the slab, with sinking of the blocks into the clay shale pressures self-generated by movement itself may play a
deposit. not negligible role, at least in the case of mudslides; fol-
44 Luciano Picarelli

lowing movements is affected by processes of pore pres- Crescenti U, Dramis F, Prestininzi A, Sorriso-Valvo M (1994) Deep-
sure equalization. The displacement rate is then a func- seated gravitational slope deformations and large-scale landslides
in Italy. A regional guide. Special volume for the International
tion of the rate of stress change, of the rate of excess pore
Congress of the IAEG, Lisboa, pp 710
pressure dissipation and of soil properties, possibly in- Cruden DM, Varnes DJ (1996) Landslide types and processes. In:
cluding a time-dependent component. Turner AK, Schuster RL (eds) Landslides: investigation and miti-
The mechanics of lateral spread is more complicated. gation. Special Report 247, Transportation research board, US
In some cases, deformation induced by erosion and, pos- National research council, Washington, D.C., pp 3675
DElia B, Picarelli L, Leroueil S, Vaunat J (1998) Geotechnical charac-
sibly, by earthquakes, is prominent. However, once again,
terisation of slope movements in structurally complex clay soils
excess pore pressures can play a significant role, mainly and stiff jointed clays. Rivista Italiana di Geotecnica 32(3):547
when the movement involves a thick deposit of highly Di Maio C (1996) The influence of pore fluid composition on the
plastic clay: in such a case, excess pore pressures can be residual shear strength of some natural clayey soils. In: Senneset
either negative, as a consequence of erosion, or positive K (ed) Proceedings 7th International Symposium on Landslides,
due to seismic input. Pore pressure equalization (consoli- Trondheim, Balkema, Rotterdam, 2, pp 11891194
Di Maio C, Onorati R (2000) Influence of pore liquid composition
dation or swelling) may strongly affect soil movement. on the shear strength of an active clay. In: Bromhead E, Dixon N,
Ibsen ML (eds) Proceedings 8th International Symposium on
Landslides. Cardiff, 1, pp 463468
References Di Nocera S, Fenelli GB, Iaccarino G, Pellegrino A, Picarelli L, Urciuoli
G (1995) An example of the geotechnical implications of geo-
Angeli MG, Gasparetto P, Menotti RM, Pasuto G, Silvano S (1996) A logical history. In: Proceedings 11th ECSMFE, The interplay be-
visco-plastic model for slope analysis applied to a mudslide in tween Geotechnical Engineering and Engineering Geology,
Cortina dAmpezzo, Italy. Q J Eng Geol 29:233240 Copenhagen, 8, pp 3948
Bertini T, Cugusi F, DElia B, Rossi-Doria M (1986) Lenti movimenti Fenelli GB (1986) Alcuni peculiari aspetti relativi alla stabilit del
di versante nellAbruzzo adriatico: caratteri e criteri di stabilizza- colle di Bisaccia. In: Proceedings 16th National Geotechnical Con-
zione. In: Proceedings 16th National Geotechnical Conference La ference, Bologna, CLEUP, Roma, 3, pp 191196
Progettazione Geotecnica per la Stabilizzazione dei Pendii, Bo- Fenelli GB, Picarelli L (1990) The pore pressure field built up in a
logna, CLEUP, Rome, 1, pp 91100 rapidly eroded soil mass. Can Geotech J 27:387392
Boucek B, Pardo-Praga D (1984) Fluage dune argile sur la surface Fenelli GB, Pellegrino A, Picarelli L (1986) Stability problems of old
de glissement. In: Proceedings 4th International Symposium on towns built on relict plateaux resting on clay deposits. In: Viggiani
Landslides, Toronto, 2, pp 247252 C (ed) International Symposium on Geotechnical Engineering
Canuti P, Casagli N, Garzonio CA, Vannocci P (1990) Lateral spreads for the Preservation of Monuments and Historic Sites. Napoli,
and landslide hazards in the Northern Apennine: the example of the Balkema, Rotterdam, pp 163174
Mt. Fumaiolo (Emilia-Romagna) and Chiusi della Verna (Tuscany). Fenelli GB, Picarelli L, Silvestri F (1992) Deformation process of a
In: Proceedings 6th Congress IAEG, Amsterdam, 3, pp 15251533 hill shaken by the Irpinia earthquake in 1980. In: Faccioli E, Pecker
Cartier G, Pouget P (1988) tude du comportement d'un remblai A (eds) Proceedings French-Italian Conference on Slope Stabil-
construit sur un versant instable: le remblai de Salldes (Puy-de- ity in Seismic Areas. Bordighera, Ouest ditions pp 4762
Dme). Rapport de recherche LPC n153, Paris Hoek E, Brown CT (1980) Empirical strength criterion for rock
Cavalera L (1977) Consolidazione per variazione periodica delle masses. J Geotech Eng-ASCE, 106(GT9):10131035
pressioni interstiziali al contorno. Rivista Italiana di Geotecnica Hungr O, Morgenstern NR (1983) High velocity ring shear tests on
11(4):187205 sand. Gotechnique 34(3)415421
Chelli A, Mandrone G, Trufelli G (2006) Field investigations and Hutchinson JN (1970) A coastal mudflow on the London clay cliffs at
monitoring as tools for modelling the Rossena castle landslide Beltinge, North Kent. Gotechnique 20(4):412438
(Northern Appennines, Italy). Landslides 3(3):252259 Hutchinson JN (1988) General report: Morphological and geotechnical
Comegna L (2005) Propriet e comportamento delle colate in argilla. parameters of landslides in relation to geology and hydrogeology.
PhD Thesis, Seconda Universit degli Studi di Napoli In: Bonnard C (ed) Proceedings 5th International Symposium on
Comegna L, Urciuoli G, Picarelli L (2004) The role of pore pressures Landslides, Lausanne, Balkema, Rotterdam1, pp 335
on the mechanics of mudslides. In: Lacerda W (ed) Proceedings Hutchinson JN Bandhari RK (1971) Undrained loading: a fundamen-
9th International Symposium on Landslides, Rio de Janeiro, 2, tal mechanism of mudflows and other mass movements. Go-
pp 11411176 technique 21(4):353358
Comegna L, Picarelli L, Urciuoli G (2007) The mechanics of mudslides Jappelli R, Musso A, Umilt G (1977) An observational approach to a
as a cyclic undrained-drained process. Landslides (accepted) creep problem in debris due to reservoir filling. In: Proceedings
Corominas J (1995) Evidence of basal erosion and shearing as mecha- International Symposium The Geotechnics of Structurally Com-
nism contributing to the development of lateral ridges in plex Formations, Capri, 2, pp 115121
mudslides, flow-slides, and other flow-like gravitational move- Kenney TC (1967) The influence of mineral composition on the re-
ments. Eng Geol 39:4570 sidual strength of natural soils. In: Proceedings Geotechnical
Corominas J, Moya J, Ledesma A, Lloret A, Gili JA (2005) Prediction Conference, Oslo, 1, pp 123129
of ground displacements and velocities from groundwater level Kenney TC, Lau KC (1984) Temporal changes of groundwater
changes at the Vallcebre landslide (Eastern Pyrenees, Spain). pressure in a natural slope of non fissured clay. Can Geotech J
Landslides 2(2):8396 21:138146
Cotecchia V, Del Prete M, Federico A, Fenelli GB, Pellegrino A, Picarelli Lampitiello S, Olivares L, Silvestri F (2001) Numerical simulation of
L (1984) Some observations on a typical mudslide in a highly tec- seismic and post-seismic response of Bisaccia Hill. In: Proceed-
tonized formation in Southern Apennines. In: Proceedings 4th In- ings TC4 Satellite Conference on Lesson Learned on Recent
ternational Symposium on Landslides, Toronto, 2, pp 3944 Strong Earthquakes, Istanbul
Chapter 3 Considerations about the Mechanics of Slow Active Landslides in Clay 45

Lembo-Fazio A, Manfredini G, Ribacchi R, Sciotti M (1984) Slope Picarelli L, Olivares L, Di Maio C, Silvestri F, Di Nocera S, Urciuoli G
failure and cliff instability in the Orvieto Hill. In: Proceedings (2002) Structure, properties and mechanical behaviour of the
4 th International Symposium on Landslides, Toronto, 1, highly plastic intensely fissured Bisaccia Clay Shale. In: Tan TS,
pp 155178 Phoon KK, Hight DW, Leroueil S (eds) Proceedings International
Leroueil S, Vaunat J, Picarelli L, Locat J, Faure R (1996) Geotechnical Symposium on Characterisation and Engineering Properties of
characterization of slope movements. In: Senneset K (ed) Pro- Natural Soils, Singapore, 2, pp 947962
ceedings 7th International Symposium on Landslides, Trondheim, Picarelli L, Urciuoli G, Russo C (2004) The role of groundwater re-
Balkema, Rotterdam,1, pp 5374 gime on behaviour of clayey slopes. Can Geotech J 41(3):467484
Mandolini A, Urciuoli G (1999) Previsione dellevoluzione cinematica Picarelli L, Urciuoli G, Ramondini M, Comegna L (2005) Main fea-
dei pendii mediante un procedimento di simulazione statistica. tures of mudslides in tectonised highly fissured clay shales. Land-
Rivista Italiana di Geotecnica 33(1):3744 slides 2(1):1530
Moore R, Brunsden D (1996) Physical-chemical effects on the Picarelli L, Urciuoli G, Mandolini A, Ramondini M (2006) Softening
behaviour of a coastal mudslide. Gotechnique 46(2):259278 and instability of natural slopes in highly fissured plastic clay
Morgenstern N, Tchalenko JS (1967) Microscopic structures in kaolin shales. Natural Hazards and Earth System Sciences 6(4):529539
subjected to direct shear direct shear. Gotechnique 17:309328 Radbruch-Hall D, Varnes DJ, Savage WZ (1976) Gravitational spread-
Nakamura H (1984) Landslides in silts and sands mainly in Japan. ing of steep-sided ridges (sackung) in western United States. Bull
In: Proceedings 4th International Symposium on Landslides, IAEG 14:2335
Toronto, 1, pp 155178 Russo C (1997) Caratteri evolutivi dei movimenti traslativi e loro
Neuzil CE (1993) Low fluid pressure within the Pierre shale: a tran- interpretazione meccanica attraverso lanalisi numerica. PhD
sient response to erosion. Water Resour Res 29(7):20072020 Thesis, Universit di Napoli Federico II
Olivares L (1997) Caratterizzazione dellargilla di Bisaccia in Skempton AW (1967) Some observations on tectonic shear zones.
condizioni monotone, cicliche e dinamiche e riflessi sul In: Proceedings II International Conference on Rock Mech.,
comportamento del Colle a seguito del terremoto del 1980. PhD Belgrade, 1, pp 329335
Thesis, Universit di Napoli Federico II Skempton AW (1985) Residual strength of clays in landslides, folded
Pasek K (1974) Gravitational block-type slope movements. In: Pro- strata and the laboratory. Gotechnique 35(1):318
ceedings 2nd International Congress IAEG, Sao Paulo, 2: V-PC-1 Tacher L, Bonnard C, Laloui L, Parriaux A (2005) Modelling the
Pellegrino A, Picarelli L, Urciuoli G (2004) Experiences of mudslides behaviour of a large landslide with respect to hydrogeological and
in Italy. In: Picarelli L (ed) Proceedings International Work on geomechanical parameter heterogeneity. Landslides 2(1):314
Occurrence and Mechanisms of Flow-Like Landslides in Earthfills Tika TM, Vaughan PR, Lemos LL (1996) Fast shearing of pre-exist-
and Natural Slopes, Sorrento. Patron, Bologna, pp 191206 ing shear zones in soils. Gotechnique 46(2):197233
Picarelli L (1988) Modellazione e monitoraggio di una colata in Tommasi P, Rotonda T (1995) Analysis of the deformations occur-
formazioni strutturalmente complesse. In: Cascini L (ed) Pro- ring at the edge of a rock cliff underneath a historical center. In:
ceedings Conference on Cartografia e Monitoraggio dei Movi- Fujii T (ed) Proceedings 8th International Congress on Rock
menti Franosi, Bologna, pp 119130 Mechanics, Tokyo, pp 383388
Picarelli L (1993) Structure and properties of clay shales involved in Tommasi P, Pellegrini P, Boldini D, Ribacchi R (2006) Influence of
earthflows. In: Anagnastopoulos A, Schlosser F, Kalteziotis N, rainfall on hydraulic condition and movement rate in the over-
Frank R (eds) Proceedings International Symposium on The consolidated clayey slope of the Orvieto Hill (Central Italy). Can
Geotechnical Engineering of Hard Soils-Soft Rocks, Athens, 3, Geotech J 7086
pp 20092019 Urciuoli G (2002) Strain preceding failure in infinite slopes. Int J
Picarelli L (2000) Mechanisms and rates of slope movements in fine Geomech 2(1):93112
grained soils. In: Proceedings GeoEng2000, International Confer- Urciuoli G, Picarelli L (2004) The shear strength mobilized in first-
ence on Geotechnical and Geological Engineering, Melbourne, time slides in highly overconsolidated clays. In: Jardine RJ, Potts
Technomic Publ. Co., 1, pp 16181670 DM, Higgins KG (eds) Proceedings International Conference on
Picarelli L (2001) Transition from slide to earthflow, and the reverse. Advances in Geotechnical Engineering. The Skempton Confer-
In: Proceedings TC-11 Satellite Conference on Transition from ence, Thomas Telford, 2, pp 10051018
Slide to Flow-Mechanisms and Remedial Messures, Trabzon, in Urciuoli G, Picarelli L, Leroueil S (2007) Local soil failure before gen-
CD-ROM eral slope failure. Geotech Geol Eng 25(1):103123
Picarelli L, Russo C (2004) Mechanics of slow active landslides and Van Genuchten PMB (1984) Intermittent sliding of a landslide in varved
interaction with man-made works. In: Lacerda W (ed) Proceed- clays. In: Bonnard C (ed) Proceedings 5th International Symposium
ings 9th International Symposium on Landslides, Rio de Janeiro, on Landslides, Lausanne, Balkema, Rotterdam,1, pp 471476
2, pp 11411176 Varley PM, Warren CD (1995) The Channel tunnel project. In: Pro-
Picarelli L, Simonelli AL (1991) Monitoring of a slow movement in ceedings 11th ECSMFE, Copenhagen, 10, pp 71117
colluvial soils. In: M. Durand M, Fabre D (eds) Proceedings 4me Vulliet L (1986) Modlisation des pentes naturelles en mouvement.
Entretiens du Centre Jacques Cartier, Colloque 4: Gotechnique PhD Thesis no 635. cole Politechnique Fdrale de Lausanne
et Environnement, Grnoble Vulliet L, Hutter K (1988) Viscous-type sliding laws for landslides.
Picarelli L, Urciuoli G (1988) Influenza della velocit di scorrimento Can Geotech J 25(3):467477
sulla resistenza residua delle argille. Gruppo Naz. di Coordin. per Wilson ST (1969) Observational data on ground movements related
gli Studi di Ingegneria Geotecnica, Attivit di ricerca nellanno to slope instability. J Soil Mech Found Div-ASCE p 96
198788, pp 101104 Yoshida N (1990) Time-dependent instability in fissured overcon-
Picarelli L, Urciuoli G (1993) Effetti dellerosione in argilliti di alta solidated clays and mudstones. PhD Thesis, Univ. of Alberta,
plasticit. Rivista Italiana di Geotecnica 17:2947 Edmonton, Canada
Chapter 4

Dynamics of Rapid Landslides

Oldrich Hungr

Abstract. Velocity is the most important parameter determining the locities of the order of 5 m s1, stand out in terms of danger
destructive potential of landslides. Catastrophic velocities of the to human life. Slower movements usually cause only mate-
order of several meters per second are attained only by certain types
rial damage. The limit of 5 m s1 corresponds to the speed
of landslides. High velocities are the consequence of a range of
strength loss mechanisms, reviewed in this chapter. Strength loss of a person running and could be called catastrophic ve-
can occur instantly during the process of failure, through loss of locity. Using terminology suggested by Varnes (1978) and
cohesion, liquefaction of granular material or remoulding of sensi- Cruden and Varnes (1996), landslides capable of reaching
tive clay. Further important loss of strength can occur during move- catastrophic velocities classify as extremely rapid.
ment, including rock joint roughness reduction, shearing in clays,
From the Work-Energy Theorem, the mean kinetic
sliding surface liquefaction, frictional heating, loss of internal co-
herence of the sliding body, material entrainment, rapid undrained energy of a landslide with a mass M can be calculated as
loading, and entrainment of water. Extremely rapid landslides in- the work of the net driving force (gravity force resultant, G,
clude rock, debris and earth fall, rock block topple, rock slide, debris minus resisting force, R, derived from material strength),
slide, flow slide in granular soil or clay, debris avalanche, debris flow moving along the landslide path, s:
and rock avalanche. There is a need to study the post-failure behav-
ior of materials, in order to facilitate predictions of the behavior of
extremely rapid landslides for hazard assessment. (4.1)
Keywords. Velocity, rapid landslide, dynamics, liquefaction, shear
strength Thus, landslide velocity is primarily the consequence
of strength loss, i.e. the decrease of R, relative to G that is
a function of slope only. This chapter reviews the causes
of strength losses that are responsible for generating cata-
4.1 Introduction strophic landslide velocities. Examples of extremely rapid
landslides are presented.
The destructive potential of a landslide depends on ve-
locity. This can be seen in published examples of land-
slide incidents, where both the typical velocity of move- 4.2 Mechanisms Causing Strength Loss in Landslides
ment and the corresponding human response were docu-
mented. A compilation of such cases by Hungr (1981) was Loss of strength can occur instantaneously as part of a
later paraphrased by the IUGS Working Group on Land- brittle failure of the soil or rock material, or during dis-
slides (1995), as seen in Table 4.1. In a summary, only land- placement as a results of changes in friction and pore-
slides in the first and second response class, exhibiting ve- pressure along the rupture surface.
48 Oldrich Hungr

been studied in the laboratory for many decades (e.g.,

4.2.1 Instantaneous Strength Loss Mechanisms
Castro 1975), usually with the help of a triaxial testing
apparatus. Numerous conditions influence the amount Loss of Cohesion in Rocks and Cemented
of strength loss, including: density of the soil, grain-size
or Unsaturated Soils
distribution, stress history, stress level and failure stress
path (e.g., McRoberts and Sladen 1985). In recent years, it
The true cohesion of granular soils cemented by calcium was realized that liquefaction affects not only sands, but a
carbonate, iron oxide or similar will disappear instanta- variety of well-graded materials, including those that con-
neously, as connected grains break apart under excessive tain significant coarse fractions (e.g., Dawson et al. 1998).
shear or tensile stress. The same may occur in partially A serious disadvantage of the triaxial test is that it is
saturated soils as suction elements at particle contacts are limited to very small displacements. The sophisticated
severed by excessive stress. The brittle initial failure of a ring shear tests reported by Sassa and his co-workers (e.g.,
silt column shown in Fig. 4.4, later in this chapter, was Sassa et al. 2004), allow the liquefaction process to be fol-
caused by a combination of cement and suction cohesion lowed to large displacements. An example is shown in
loss. Root cohesion in shallow soils can also be destroyed Fig. 4.1, presenting the time series and stress paths of an
suddenly by overstress, albeit with a few cm of displace- undrained cyclic ring shear test on loose sand. The sand
ment. Cohesion of rock masses results largely from the was collected from the vicinity of the rupture surface of
presence of intact rock bridges, separating non-persis- an extremely rapid landslide, triggered by the 2004 Mid-
tent joints. (e.g., Eberhardt et al. 2004). When subjected Niigata Prefecture earthquake (Sassa et al. 2005). The ring
to excessive stress, the bridges are destroyed through shear shear test was conducted by transferring simulated dy-
or crack propagation, often with a dramatic loss of namic normal and shear stress signals into the annular
strength combined with very small displacement. test box (green and red traces in Fig. 4.1a). As the sample
deformed under the cyclic load, pore-pressure steadily Liquefaction of Granular Soils increased (blue line). Initial pore-pressure increase of a
few percent of the normal stress occurred with very small
Spontaneous or earthquake liquefaction refers to a col- displacement. This phenomenon could be compared to
lapse of a loose soil skeleton due to excessive strain, fol- the cyclic mobility observed frequently in triaxial tests.
lowed by an increase in pore-pressure and loss of strength More severe pore-pressure increase occurred with in-
under undrained conditions. The liquefaction process has creasing displacement. After 50 mm of movement, the
pore pressure ratio within the box reached a level of ap-
proximately 0.75, indicating that a high degree of lique-
faction has occurred. The displacement of 50 mm is about
twice as large as can be achieved in a typical triaxial test.
Additional pore-pressure generation followed this initial
result, as discussed below.
Recent studies of earthquake liquefaction showed that
the undrained effective stress response in soils of non-
uniform permeability can be very different from that
which would take place in a homogeneous soil. This is an
effect that cannot easily be studied in laboratory tests
(Byrne et al. 2006).
The above considerations and the fact that many land-
slides involve heterogeneous soils, indicate that the re-
sults of laboratory testing must be interpreted with cau-
tion, if they are to provide constitutive relationships for
use in quantitative analyses of flow slides. This is true
especially of the traditional undrained triaxial laboratory
test with limited displacement. Remoulding of Sensitive Clays

Certain clays of marine origin, such as the quick clays of

north-eastern North America and southern Scandinavia are
Fig. 4.1. Undrained cyclic ring shear test on sand from the Higashi- characterized by extreme loss of strength on remoulding
Takezawa landslide (from Sassa et al. 2005) following shear failure (e.g., Crawford 1968). The rheo-
Chapter 4 Dynamics of Rapid Landslides 49

logical properties of remoulded extra-sensitive clay can be is capable of denser packing. This generates pore pres-
those of a Newtonian viscous fluid, with negligible shear sure under undrained conditions and a corresponding loss
strengths. Unfortunately, laboratory-derived viscosity of strength. The effect can be observed to some degree in
values cannot be used directly to analyze full-scale flows, Fig. 4.1. The initial structural collapse of the soil skeleton
due to a scale-dependence of the rheology (Locat 1993). reduced the effective stress by some 75% after 50 mm of
displacement. Following this, however, a further gradual
reduction of about 15% occurred, to reach a steady-state
4.2.2 Strength Loss Requiring Large Displacements pore-pressure ratio of about 0.9 after 200 mm of displace-
ment. Fukuoka et al. (2006) showed that measurable Surface Roughness Reduction changes in grain size distribution in the vicinity of the
in Rock Discontinuities sliding surface follow such tests. The phenomenon ap-
pears to be most prominent in soils with relatively weak
Large shearing displacements applied to rock disconti- grains, especially pumice.
nuities will reduce the dilatancy component of frictional
strength, producing residual friction (e.g., Hoek and Bray Frictional Heating
1974). For example, Cruden and Krahn (1978) tested the
peak friction angle of joints and bedding planes in the Friction across a sliding surface produces heat. The coef-
limestone of Turtle Mountain (the Frank Slide) as ranging ficient of volumetric thermal expansion of water is about
between 32 and 52 with small cohesion, while the ultimate an order of magnitude larger than that of mineral par-
friction of friction-polished joints was 14 to 32. No labora- ticles. As a result, undrained heating of saturated soil will
tory tests have yet been carried out that would approximate be accompanied by an increase in pore pressure, hence a
the effects of long-displacement shearing at high normal reduction of shear strength. Such effects have been shown
stresses, such as must occur at the base of thick transla- theoretically by Voight and Faust (1982). Earlier, Hendron
tional or compound slides in rock. Therefore, little is known and Patton (1985) proposed that frictional heating led to
about the amount of displacement needed to produce true the reduction of shear strength at the base of the Vaiont
ultimate strength of natural rock joints. The effects of Slide. This is a plausible idea for certain large rock slides
scale will undoubtedly contribute to the interpretation of although, to the Authors knowledge, no experimental or
such tests. Nevertheless, for runout analyses of large rock field confirmation exists to date.
slides, a substantial reduction of the friction angle fol-
lowing a few meters of displacement is to be expected. Loss of Internal Coherence of the Sliding Body Shearing in Clays The resisting forces in a slide derive not only from the
strength of the rupture surface, but also to some extent
Frictional strength of clay is also known to decrease after from the mobilized internal strength of the sliding body.
long shear displacements, as a result of reorientation of the Therefore, the internal brittleness of a rigid slide may in
platy clay particles and the development of a sheared struc- some cases add to the brittleness of the sliding surface.
ture (Skempton 1985). The amount of decrease (brittle- Hutchinson (1988) proposed that the rapid acceleration
ness) depends on the type of clay mineral and the clay con- of the Vaiont Slide was partly due to the brittle loss of
tent and can amount to more than 50%. In the laboratory, cohesion within the limestone mass, sliding on a clay-
the residual strength is fully developed after only a few hun- coated bedding plane. Based on kinematic considerations,
dreds of mm. The corresponding minimum displacement the mobilization of internal strength requires greater dis-
in the field may be much greater. Recent laboratory testing placement than the mobilization of the boundary strength
indicated that the residual friction of certain clays may de- on the rupture surface.
crease further by as much as 60% at high rates of shearing,
increasing brittleness (Tika and Hutchinson 1999). Material Entrainment and Rapid Undrained
Loading Sliding Surface Liquefaction
The rapid undrained loading mechanism was docu-
The concept of sliding surface liquefaction in granular mented by Hutchinson and Bhandari (1971), who ob-
material, proposed by Sassa (2000) is somewhat analo- served pore pressure increases in a piezometer embed-
gous to the formation of residual strength in clays. Once ded near the basal surface of an earth flow subjected to
failure occurs and is followed by large displacement, the rapid deposition of material from upslope. This mecha-
intense shearing within a thin shear band leads to a tex- nism appears to be the principal means of mobilizing
tural change within the material (grain crushing). The earth flow movements on gentle slopes. Sassa (1985) ex-
modified soil contains a greater proportion of fines and tended the same principle to debris avalanches and de-
50 Oldrich Hungr

Fig. 4.2. Time series and stress path plots of rapid undrained load- Fig. 4.3. Process of simultaneous entrainment, plowing and lubri-
ing of a sample from a debris flow in a weathered volcanic soil cation at the front of a rock avalanche, described by Hungr and
(Sassa et al. 2004) Evans (2004)

bris flows. It has long been known that debris flows de-
4.3 Types of Extremely Rapid Landslides
rive most of their volume by entraining loose saturated
material from their path (e.g., Hungr et al. 2005). Such
material is in all probability mobilized by rapid undrained A simple form of the popular Varnes (1978) classification of
loading, caused by debris surges arriving from upstream. landslides can be represented by 26 keywords, each repre-
Figure 4.2 shows the results of an undrained ring shear senting a unique type of slope movement. The keywords are
test under sudden loading by Sassa et al. (2005), illustrat- shown in Table 4.2 within Varnes two-dimensional classifi-
ing this phenomenon. cation framework. A letter type code is used in the table to
The entrainment mechanism has also been described distinguish phenomena considered capable of producing
for rock avalanches and rock slide-debris avalanches catastrophic velocities from those that do not. Some of the
(Hungr and Evans 2004). The process has two, often si- keywords, related to landslides of flow-like character, were
multaneous consequences, both of which reduce strength: later re-defined by Hungr et al. (2001), as shown in Table 4.3.
(1) It leads to lubrication of the base of the slide by liquid A number of individual landslide types capable of ex-
saturated material, and (2) it incorporates such material tremely rapid motion, chosen from a combination of the
into the body of the landslide, increasing its volume and two classification systems shown in Tables 4.2 and 4.3 are
internal fluidity (Fig. 4.3). described as follows. Entrainment of Water, Dilution Rock Fall (fragmental rock fall, cf. Evans and Hungr 1993)
is the free fall, bouncing and rolling of rock fragments
Rapidly moving landslides also incorporate surface wa- down a slope. The initial acceleration results usually from
ter flowing in the path. This is especially true of confined loss of cohesion. Subsequent propagation mechanisms
debris flows that often follow natural drainage paths or involve free fall, bouncing and rolling and the movement
stream channels. Wallance (2005) gives a description of is invariably extremely rapid. The volumes are limited.
gradual dilution that can change a large volcanic debris Figure 4.4 shows a dynamic back-analysis of the fall of a
flow into a flood-like hyperconcentrated flow. Dilution is 100 tonne boulder which caused two fatalities in British
often accompanied by a gradual reduction in peak dis- Columbia in 1982 (Evans and Hungr 1993).
charge of debris surges and this leads to some decrease in
their destructive potential. On the other hand, it can Debris Fall involves similar free movement of solid frag-
greatly increase the ultimate runout distance. ments derived from coarse soil.
Chapter 4 Dynamics of Rapid Landslides 51

Earth Fall results from localized failure of fine-grained, terial in situ is a combination of suction forces in the un-
cohesive soil. High movement velocity again results from saturated soil and cementing by a trace of clay and cal-
loss of cohesion. Figure 4.5 shows the flow of dry glacio- cium carbonate cement. A column of silt, separated from
lacustrine silt following the failure of a part of a vertical the cliff face along a vertical stress-relief joint, toppled,
face, supported by cohesion. The cohesion of the silt ma- fell and disintegrated, to produce an extremely rapid flow
52 Oldrich Hungr

Fig. 4.4. Dynamic analysis of a 100 tonne rock fall. a Damage, b boulder (Hungr and Evans 1985)

Fig. 4.5. Dry silt flow near Peachland, British Columbia (photo courtesy of Mr. J. Valentinuzzi, British Columbia Ministry of Transportation
and Highways)

that crossed a major highway and covered a distance of Rock Block Topple. Block toppling involves relatively mas-
100 m in about 8 seconds. A dynamic back-analysis of the sive blocks separated by near-vertical joints and resting
failure using the program DAN (Hungr 1995), confirmed on structurally-defined basal surfaces. Each block origi-
that the disintegrated silt was dry and frictional, travel- nally derives its stability from the position of the center
ing with a bulk friction angle of 30, corresponding to the of gravity with respect to the base of the block. The block
dynamic friction angle of dry silt. The high velocity in rotates forward gradually under the influence of some
this case was thus due solely to the release of potential destabilizing agent (e.g., water or ice pressure in the ten-
energy as a result of cohesion loss in course of the initial sion crack). Once the point of overturning is reached,
failure. No pore-pressure was involved. however, further rotation becomes irreversible and ex-
Chapter 4 Dynamics of Rapid Landslides 53

Fig. 4.6.
A block topple, Czech Republic

tremely rapid. Block toppling is the only type of slope Vaiont, Hendron and Patton 1985), but more often disin-
failure where high velocity does not require strength loss, tegrates to produce a rock avalanche (Fig. 4.7).
but results from the inherent instability of the failure The presence of non-persistent discontinuities, sepa-
mechanism. Block toppling mechanism affects single rated by bridges of intact rock, contribute a cohesion,
blocks, small groups of blocks (see Goodman and Bray the loss of which usually leads to dramatic acceleration.
1976 and Fig. 4.6) and possibly also certain large rock Hungr and Evans (2004a) referred to this type of sliding
masses (e.g., Chigira and Kiho 1994). mechanism as a rock collapse.

Rock Slide. Sliding failure in strong rock requires discon- Debris Slide. Debris slides in granular material involve
tinuities. Where the discontinuities are persistent, failure translational sliding of a thin colluvial or residual veneer
brittleness will result from surface roughness reduction, over strong substrate (e.g., Savage and Baum 2005). Slow-
as discussed above. It is probably for this reason that struc- moving debris slides exist, but are common only in rela-
turally-controlled translational and compound sliding in tively dry loose soils (for example, slip faces of sand
strong rocks produces highly mobile rock slides. The ini- dunes). Usually, strength losses in the form of cohesion
tial sliding block sometimes remains semi-coherent (e.g., loss and full or partial liquefaction combine to produce
54 Oldrich Hungr

Fig. 4.7.
Translational failure of sedi-
mentary sequence of siltstone
and sandstone, Rockslide Pass,
Mackenzie Mountains, Canada.
The landslide began by sliding
of a 600 m thick block on a bed-
ding plane. After colliding with
the valley fill to the right of the
center of the photo, the block
disintegrated and flowed for
another 4 km as a rock ava-
lanche (Hungr and Evans 2004)

acceleration, while rapid undrained loading or sliding

surface liquefaction (Sassa 2000) lead to further increase
of speed, disintegration of the granular mass and the for-
mation of an extremely rapid debris avalanche.

Flow Slide in Granular Soil. The classical concept of flow slid-

ing, defined by Casagrande (1976) and others, involves
extremely rapid, flow-like movement following sudden
strength loss due to spontaneous or earthquake liquefac-
tion of loose, saturated granular soil. The process takes
place at constant volume, with no increase in water content.
Many flow slides occur in poorly consolidated sandy sedi-
ments under water. Older accounts of flow slides describe
very loose sands or silty sands. More recently, however, it Fig. 4.8. A flow slide in coal mining waste, South-east British Colum-
was realized that liquefaction can involve a wide variety of bia, Canada. The 60 000 m3 slide originated on the waste pile in the
granular materials, including those that are relatively coarse background and traveled 2 km
and well-graded such as colluvium and mine waste (e.g.,
Dawson et al. 1998; Hunter and Fell 2002, see Fig. 4.8). mass of earth supported by a failing sensitive layer (cf.
While saturation is key to the basic liquefaction mecha- flake slide, Hutchinson 1988). The liquefied clay flows
nism, this does not mean that liquefaction flow slides are out of a slide crater and follows the slope, often along a
limited to completely saturated slopes. In many cases, the river channel. The flow is that of a Newtonian viscous
presence of spontaneous or earthquake liquefaction can substance, often in the turbulent regime (Locat 1993) The
only be deduced from the sudden acceleration of the ini- flow carries intact floes of broken desiccated surficial
tial failure and rapidity and flow-like character of the sub- crust, that can influence the flow behavior (Hungr 1981).
sequent motion. The bulk of the soil or rock may be in an
unsaturated condition, as long as sufficient saturation Debris Avalanche. As described earlier, a debris avalanche
exists in the vicinity of the rupture surface. is the end result of shallow sliding of granular material
on steep hillslopes (Fig. 4.9). The initial slide is metamor-
Flow Slide in Extra-Sensitive Clay. A slide in an extra-sensi- phosed into a rapid flow through a combination of
tive clay, often on a river bank undermined by erosion, strength loss mechanisms, that may include cohesion loss,
produces dramatic loss of strength by remoulding. The spontaneous liquefaction, rapid loading of material in the
mobile debris evacuates the source area and a steep scarp path and sliding-surface liquefaction. In conducting dy-
at the current slide head becomes exposed to further fail- namic analysis it is exceedingly difficult to separate the
ure. The process may take the form of multiple retrogres- effects of these various contributing effects. The concept
sive rotational slides, or a spontaneous displacement of a of equivalent fluid, introduced by Hungr (1995) avoids
Chapter 4 Dynamics of Rapid Landslides 55

this complex issue, by assuming that the solid landslide The boundary between flow slides, dominated by liq-
material in the source area turns instantly into a simple uefaction, and debris avalanches that are mobilized by
fluid, the characteristics of which must be determined by several simultaneous strength loss processes, is difficult
back analysis. to define. A well-known case of an extremely rapid land-
slide from Hong Kong illustrates the problem:
The Shum-Wan Road landslide took place on Au-
gust 13, 1995 (Fig. 4.10). Failure began at the end of a rain
storm with total precipitation of 159 mm and hourly in-
tensity of up to 48 mm, preceded by 846 mm of rain dur-
ing the preceding 13 days. As interpreted by Hong Kong
Geotechnical Engineering Office (GEO 2006), the landslide
began by the failure of a road fill, coinciding with the re-
lease of large amounts of surface water onto the slope be-
low Shum Wan Road. Immediately after the initial failure, a
spoon-shaped rupture surface developed in the slope made
up of volcanic residual soil, utilizing relict joints and seams
filled with kaolinite (Fig. 4.11). The main body of the land-
slide accelerated to a high speed, mobilizing a thin slab of
residual soil further downslope. The whole landslide of
26 000 m3 came to a stop within a few minutes of the initial
movement, having covered some 80 m distance at the toe.
A dynamic analysis of the landslide using a bulk fric-
tion coefficient of 20 (including pore pressure effects)
yielded good correspondence with the observed move-
ment behavior (Ayotte et al. 1999). In contrast, static slope
stability analysis used several surfaces with friction angles
between 26 and 38 and small cohesion. The fall of strength
responsible for the rapid movement may have resulted
partly from cohesion loss, from shearing of kaolinite from
peak to residual friction angle (26 to 21) and from un-
drained loading. Both the fill and colluviated surficial
Fig. 4.9. A debris avalanche in colluvium, Jasper National Park, Ca- residual soil were probably susceptible to liquefaction.
nadian Rocky Mountains. These diverse sources of potential strength loss cannot

Fig. 4.10.
The Shum Wan Road debris
avalanche, Hong Kong (photo
courtesy of Geotechnical Engi-
neering Office, Hong Kong)
56 Oldrich Hungr

Fig. 4.11. Profile of the Shun Wan Road debris avalanche (GEO 2006)

Rock Avalanche. A rock avalanche is the extremely rapid

flow-like movement of fragmented rock from a large rock
slide. The initial acceleration is provided by cohesion loss
and joint surface roughness reduction. In case of certain
planar detachments such as that shown on Fig. 4.7, fric-
tional heating may also play a role. When the rock ava-
lanche mass over-rides saturated surficial soils, a pro-
cess of rapid loading and material entrainment will take
place to mobilize long-runout movement (e.g., Sassa
1988; Hungr and Evans 2004). Several other physical ex-
Fig. 4.12. A schematic profile of a debris flow surge (Pierson 1986) planations of rock avalanche mobility have been ad-
vanced, but none has so far been shown to be universal
easily be separated from each other and the author pre- (e.g., Hungr 2006).
fers to refer to this occurrence as a debris avalanche.

Debris Flow. Debris flows are extremely rapid, surging flows 4.4 Conclusion
of unsorted, saturated debris in pre-defined channels
(Hungr et al. 2001). Many debris flows form from debris Formation of high velocity in landslides occurs pri-
avalanches starting on steep slopes and entering a channel. marily through strength loss. The physical proces-
The process of surge formation, conditioned by the ses leading to loss of strength in earth materials are
presence of a channel, give debris flows their extraordi- many and often they work together. Some strength
nary mobility and destructive character. In coarse debris, loss occurs instantaneously at the point of failure, other
the surges result from longitudinal sorting. A concentra- involves gradual change of material during motion.
tion of coarse clasts forms at the surge front, followed by Existing research on slope stability and failure behavior
increasingly more fine-grained and dilute debris (Pierson has concentrated almost exclusively on processes
1986, Fig. 4.12). As a result, the peak depth and discharge leading to instability. In order to predict landslide be-
of the surge increase in course of movement down the havior and runout for hazard assessment, it is neces-
channel. The physical properties of the material forming sary to study the mechanisms that determine post-
the surge are heterogeneous and velocity of the surge failure loss of strength. One step in the right direc-
movement is determined by the composition and water tion is the use of long-displacement testing methods,
content of material at the surge peak (Hungr 2000). Ma- advanced in recent decades by the Disaster Pre-
terial entrainment and dilution by surface water are im- vention Research Institute at Kyoto University (Sassa
portant processes controlling the dynamics of debris flow. 2000).
Chapter 4 Dynamics of Rapid Landslides 57

Hungr O, Evans SG (2004) The occurrence and classification of mas-

Acknowledgment sive rock slope failure. Felsbau 22:1623
Hungr O, Evans SG, Bovis M, Hutchinson JN (2001) Review of the
classification of landslides of the flow type. Environ Eng Geosci
The author thanks the Head of the Geotechnical Engi- VII:221238
neering Office and the Director of Civil Engineering and Hungr O, McDougall S, Bovis M (2005) Entrainment of material by
Development, Hong Kong SAR Government, for the per- debris flows. In: Jakob M, Hungr O (eds) Debris flow hazards and
mission to publish the Figs. 3.10 and 3.11. related phenomena. Springer Verlag, Heidelberg, in association
with Praxis Publishing Ltd, Chapter 7, pp 135158
Hunter GJ, Fell R (1992) Mechanics of failure of soil slopes leading
to rapid failure. In: Picarelli L (ed) Fast slope movements pre-
References diction and prevention for risk mitigation. Patron Editore, Bolo-
gna, 1, pp 283290
Ayotte D, Evans N, Hungr O (1999) Runout analysis of debris flows Hutchinson JN (1988) General report: Morphological and geotechnical
and avalanches in Hong Kong. In: Proceedings Slope Stability parameters of landslides in relation to geology and hydrogeology.
and Landslides, Vancouver, Geotechnical Society Symposium, In: Bonnard C (ed) Proceedings 5th International Symposium on
May 1999, pp 3946 Landslides, Lausanne, Balkema, Rotterdam1, pp 335
Byrne PM, Naesgaard E, Seid-Karbasi M (2006) Analysis and design Hutchinson JN, Bhandari RK (1971) Undrained loading: a funda-
of earth structures to resist seismic soil liquefaction. Keynote mental mechanism of mudflows and other mass movements.
Paper, 59th Canadian Geotechnical Conference, Canadian Geo- Gotechnique 21(4):353358
technical Society, Vancouver IUGS Working Group on Landslides (1995) A suggested method
Casagrande A (1976) Liquefaction and cyclic deformation of sands; for describing the rate of movement of a landslide. IAEG Bull
a critical review. Harvard Soil Mechanics Series, No 88, p 51 52:7578
Castro G (1975) Liquefaction and cyclic deformation of sands. J Locat J (1993) Fra fjell til fjord: consideration on viscous flows. In:
Geotech Eng-ASCE 101:551569 Proceedings, Pierre Beghin International Workshop on Rapid
Chigira M, Kiho K (1994) Deep-seated rockslide-avalanches preceded Gravitational Movements, CEMAGFEF, Grenoble University Press,
by mass rock creep of sedimentary rocks in the Kaishi Moun- pp 197207
tains, central Japan. Eng Geol 38:221230 McRoberts ED, Sladen JA (1985) Observations on static and earth-
Crawford CB (1968) Quick clays of Eastern Canada. Eng Geol 2:239265 quake liquefaction methodologies. In: Proceedings 43rd Canadian
Cruden DM, Krahn J (1978) Frank rockslide, Alberta, Canada. In: Voight Geotechnical Conference, Quebec City, Canadian Geotechnical
B (ed) Rockslides and avalanches, Vol. 1. Elsevier, Amster-dam, Society, 1, pp 215226
pp 97112 Pierson T (1986) Flow behaviour of channelized debris flows, Mt. St.
Cruden DM, Varnes DJ (1996) Landslide types and processes. In: Helens. In: Abrahams AD (ed) Hillslope processes. Allen and
Turner AK, Schuster RL (eds) Landslides: investigation and miti- Unwin, Boston, pp 269296
gation. Special Report 247, Transportation research board, US Sassa K (1985) The mechanism of debris flows. In: Proceedings of
National research council, Washington, D.C., pp 3675 XI International Conference on Soil Mechanics and Foundation
Dawson RF, Morgenstern NR, Stokes A (1998) Liquefaction flow slides in Engineering, San Francisco, 3, pp 11731176
Rocky Mountain coal mine waste dumps. Can Geotech J 35:328343 Sassa K (1988) Geotechnical model for the motion of landslides. In:
Eberhardt E, Stead D, Coggan JS (2004) Numerical analysis of initia- Special Lecture of 5th International Symposium on Landslides,
tion and progressive failure in natural rock slopes the 1991 Landslides, 1015 July, 1, pp 3755
Randa rockslide. International Journal of Rock Mech and Min- Sassa K (2000) Mechanism of flows in granular soils. In: Proceed-
ing Sciences 41(1):6987 ings of the International Conference of Geotechnical and Geo-
Evans SG, Hungr O (1993) The assessment of rockfall hazards at the logical Engineering, GEOENG2000, Melbourne, 1, pp 16711702
base of talus slopes. Can Geotech J 30:620636 Sassa K, Fukuoka H, Wang GH, Ishikawa N (2004) Undrained dy-
Fukuoka H, Sassa K, Wang G, Sasaki R (2006) Observation of shear namic-loading ring-shear apparatus and its application to land-
development in ring-shear apparatus with a transparent shear slide dynamics. Landslides 1:719
box. Landslides 3:239251 Sassa K, Fukuoka H, Wang FW, Wang G (2005) Dynamic properties
Geotechnical Engineering Office (2006) Report on the Shum Wan of earthquake-induced large-scale rapid landslides within past
Road landslide of 13 August, 1995. GEO Report No. 178, Govern- landslide masses. Landslides 2(2):125134
ment of Hong Kong, 115p Savage W, Baum R (2005) Instability of steep slopes. In: Jakob M,
Goodman RE, Bray JW (1976) Toppling of rock slopes. Rock Engi- Hungr O (eds) Debris flow hazards and related phenomena.
neering, ASCE Geotechnical Engineering Division Conference, Springer-Verlag, Berlin Heidelberg, in association with Praxis
Boulder, 2, pp 201237 Publishing Ltd, chapt. 4, pp 5377
Hendron AJ, Patton FD (1985) The Vaiont Slide. US Corps of Engi- Skempton AW (1985) Residual strength of clays in landslides, folded
neers Technical Report GL-858 strata and the laboratory. Gotechnique 35(1):318
Hoek E, Bray JW (1997) Rock slope engineerig. 2nded., Inst. Mining Tika TE, Hutchinson JN (1999) Ring shear tests on soil from the Vaiont
and Metallurgy, London landslide slip surface. Gotechnique 49(1):5974
Hungr O (1981) Dynamics of rock avalanches and other types of Vallance J (2005) Volcanic debris flows. In: Jakob M, Hungr O (eds) De-
slope movements. Ph.D. Thesis, University of Alberta, 500 p bris flow hazards and related phenomena. Springer-Verlag, Berlin
Hungr O (1995) A model for the runout analysis of rapid flow slides, Heidelberg, in association with Praxis Publishing Ltd, pp 247271
debris flows and avalanches. Can Geotech J 32(4):610623 Varnes DJ (1978) Slope movement types and processes. In: Landslides,
Hungr O (2000) Analysis of debris flow surges using the theory of analysis and control. National Academy of Sciences, National Re-
uniformly progressive flow. Earth Surf Proc Land 25:113 search Council, Washington, DC., Special Rep. 176:1133
Hungr O (2006) Rock avalanche occurrence, process and modelling. Voight B, Faust C (1982) Frictional heat and strength loss in some
Springer NATO Science Series, IV. Earth Environ Sci 49:243266 rapid landslides. Gotechnique 32:4354
Chapter 5

Progress in Debris Flow Modeling

Tamotsu Takahashi

Abstract. The processes of the two major types of debris flow initia- mixing of debris with water that has been contained origi-
tion; the bed erosion type and the landslide-induced type, are dis- nally in the earth mass or supplied from behind. The third
cussed and the methods to analyze the respective phenomena are
type is the destruction of natural dam. A landslide some-
introduced. Especially for the landslide-induced type, a new model
of liquefaction of the slid earth mass while in motion is introduced. times dams up a river, and sooner or later, the dam is de-
The earth mass is liquefied at the bottom without water supply from stroyed by the overtopping of river water or by the col-
the outside and this liquefied layer gets behind the mass as a fol- lapse of the dam body itself under the effects of seepage
lowing debris flow. The reviews of previous investigations on the water and/or hydraulic pressure. From the mechanical
mechanics of developed debris flow confirms that the two-phase
point of view, the debris flow initiation process due to
model is influential, in which debris flow consists of the mixture of
two continuum media of fluid phase and solid phase. Using this overtopping of a natural dam can be considered similar
model, developed debris flows are classified into three types from to the first type, and that due to collapse of the dam body
the point of the dominant stresses within flow. The characteristics is akin to the second type.
of the respective types of flow such as the solids concentration dis- High fluidity of debris flow is secured by sustaining
tribution, the velocity distribution, the equilibrium sediment trans- the particles apart from each other. The resistance to flow
port concentration are given.
must be the sum of the resistance due to deformation of
Keywords. Debris flow, erosion type debris flow, landslide-induced interstitial fluid and/or turbulent mixing of particles and
debris flow, single-phase continuum model, mixture theory, consti- fluid and that due to consumption of energy to disperse
tutive relations, quasi-static debris flow, stony type, turbulent muddy the particles by inter-particle collision, and the larger the
type, viscous type, immature debris flow, inertial debris flow particle concentration, the larger the resistance to flow
becomes. The necessary energy to disperse particles is
supplied from the shearing of the mixture. In an equilib-
rium state, the resistance to flow balances with the driv-
5.1 Introduction ing force due to gravity. If debris flow plunges into a gen-
tly sloping area, the particle dispersing force becomes
Debris flow is composed of a dense mixture of water and small because the velocity and hence the shearing rate of
sediment, and it surges down as if it were comprised of a the mixture becomes small, and the particles tend to settle
kind of continuous fluid. For the initiation of debris flow, down. Then, the particles concentrate denser than before
as the solid particles in debris flow have originally been in the lower part of flow. This makes the resistance larger
the components of a stable soil mass on a gully bed or on than in the upstream channel, thereby the particle dis-
a slope, some natural forces that destroy the skeleton of persing force becomes smaller. Thus, an abrupt deposi-
the mass must operate. The condition for destruction of tion occurs at the channel slope change.
the skeleton of soil mass is common to that of landslides, The mechanism to disperse the particles in flow is dif-
but the condition for keeping the flow to surge down re- ferent depending on the properties of the materials, chan-
quire a steep slope and an ample quantity of water to fill nel slope, flow rate, particle concentration, etc., and the be-
the enlarged void space between particles. havior of flow is also diverse. Theoretical discussions on
Three types of debris flow initiation predominate. The the mechanism to disperse the particles come to a conclu-
first type is due to the erosion of channel bed. Following sion that debris flows can be classified into three typical
a severe rainfall, surface water flow appears on a steep types; the stony type, the turbulent muddy flow type and
channel bed on which plenty of debris is accumulating, viscous type debris flows. The stony and turbulent muddy
and the water flow destabilizes the bed and entrains the flow types can be grouped as the inertial debris flow.
debris rapidly to form debris flow. The second type is due In this paper the author attempts to systematize the
to landslide. The slid earth mass is successively destroyed mechanical understandings of debris flows and gives the
while in motion, and it transforms into debris flow by methods or the perspectives for the quantitative analyses.
Chapter 5 Progress in Debris Flow Modeling 61

can obtain the downstream debris flow hydrograph un-

der appropriate parameter values appeared in the gov-
erning equations. Figure 5.1 is the schematic illustration
of debris flow on a uniform erodible bed under a con-
stant rate water supply from the upstream end. The steady
flow region in which no bed erosion takes place elongates
itself downstream and the erosion proceeds to the bed-
rock in the developing region that follows the steady flow
region as long as the steady water supply at the upstream Fig. 5.2. Characteristic shear distributions within the water saturated
end continues. Examples of the application of these equa- bed under the surface water flow
tions such as to obtain the hydrograph due to the col-
lapse of natural dam by overtopping and that due to flood face flow appears. This type of instability in bed is due
runoff in the actual basin can be found elsewhere (Taka- not to the dynamic force of fluid flow but to static dis-
hashi and Nakagawa 1994; Takahashi et al. 2001). Some- equilibrium, so that the flow should be called sediment
times, a shallow surface landslide transforms into debris gravity flow.
flow while it is moving on slope. Even in such a case the The condition for the occurrence of sediment gravity
analysis of debris flow development in the channel can flow is, therefore, aL dp. Substitution of this condition
be carried out without difficulty by giving the debris flow into Eq. 5.9 obtains
hydrograph and sediment concentration in it as the
boundary conditions. The prediction of that boundary
condition is, however, difficult problem to be investigated.

Even though the condition aL dp is satisfied, when aL

5.3 Criteria of Debris Flow Occurrence is far less than h0 grains cannot be uniformly dispersed
for Erosion Type throughout the whole depth of flow due to rather small
ability to disperse by the effect of particle collision but
Aforementioned discussion ignores how dense the equi- they concentrate in the lower layer of flow. Takahashi
librium solids concentration is. The analysis is common (1991) named this kind of sediment gravity flow as im-
to the cases of bed load transport and debris flow as long mature debris flow. Therefore, a sediment gravity flow that
as the appropriate f and C values are selected to the re- is appropriately called a debris flow should meet the con-
spective cases. Therefore, the criteria for occurrence of vari- dition aL kh0, in which k is a numerical coefficient, de-
ous modes in sediment transportation must be made clear. termined from experiment to be about 0.7~1. Substitu-
Imagine a thick uniform layer of loosely packed non- tion of the condition aL kh0 into Eq. 5.9 gives
cohesive grains, whose slope angle is . It is assumed that
at the moment when the surface water flow of depth h0
appears, the pore spaces among grains are saturated with
water and a parallel seepage flow without any excessive
pore pressure occurs. The characteristic distribution of Debris flow in which grains are uniformly dispersed
shear stress in the bed should be like one of those shown throughout the depth occurs when Eqs. 5.10 and 5.11 are
in Fig. 5.2, in which is the operating tangential stress simultaneously satisfied.
and L the internal resistive stress. The surface water flow on the unsaturated bed would
Case 1 in Fig. 5.2 occurs under the condition entrain grains by the effect of fluid dynamic force and
develop to debris flow. Sometimes when permeability of
the bed is very large, the infiltration of water from debris
flow into the bed retards or even stops the motion and
the accumulation of surface water flow from behind reju-
When Case 2 occurs, the following equation should be venates the debris flow motion. Thus, the motion of de-
satisfied: bris flow due to rather small surface runoff is sometimes
Attention must be paid to Case 1 where Eqs. 5.10 and
5.11 are satisfied. A grain bed that satisfies Eq. 5.8 might
slip even though the seepage flow did not reach the sur-
in which aL is the depth where and L coincide. face. In such a case, in the neighborhood of the surface
The entire bed in Case 1 and the part above the layer, L would be larger than , and there the skeletal
depth aL in Case 2 will begin to flow as soon as the sur- structure of the bed would be maintained. This phenom-
62 Tamotsu Takahashi

enon should be called landslide rather than debris flow. Consider a situation as illustrated by Fig. 5.4 on a satu-
Therefore, on such a steep permeable bed a landslide rated bed whose slope angle satisfies the condition for
would occur before the increment of the stage of seepage Case 2. The depth aL where and L coincide is given as
flow reaches the surface of the bed. The extent of motion
of such a landslide may, however, be short due to deficit (5.12)
of water. A pile of debris in a gully once moved as a land-
slide and stopped in a short range may have a smaller
In this case the denominator of Eq. 5.12 is positive and if
gradient at least on the back than it originally had on the
gully bed, so that it may be moved again as debris flow
waiting for the appearance of a surface water flow. The (5.13)
time lag between the occurrence of a landslide and the
removal of its deposit by the generation of debris flow is satisfied, aL becomes 0 meaning that the bed will not be
would depend on the water supply (rainfall) condition, eroded, and if C < C, aL > 0 and the bed will be eroded
and it sometimes could be so short that one would hardly and the grain concentration in flow will increase. There-
recognize the transition between the two phenomena. fore, Eq. 5.13 indicates the equilibrium grain concentra-
To sum up the discussion above, fully developed de- tion in debris flow on a slope angle that satisfies the con-
bris flow can occur on the bed that satisfies Eq. 5.11. dition for Case 2 but milder than the slope given by Eq. 5.8.
As shown in Fig. 5.3, the domains of occurrence of vari-
ous type sediment transportations are defined by Eqs. 5.9,
5.10, 5.11, and the equation of critical tractive force on a
steep channel (Ashida et al. 1977). The broken lines la-
beled D and E are the experimentally obtained critical
lines for the occurrence of debris flow on unsaturated bed
(Tognacca et al. 2000) and on saturated bed (Armanini
and Gregoretti 2000), respectively. The second abscissa
axis that represents the relative depth h0 /dp is attached
for the convenience to get the general idea of necessary
depth of surface flow and the scale markings on the two
abscissa axes are not rigorously correspondent. Attention
must be paid to this figure that the discussion is for the
onset of debris flow by the supply of plain surface water
flow, and once debris flow is formed it can flow on a flat- Fig. 5.4. The process of debris flow front development on a uniform
ter slope than that is necessary to generate. erodible bed

Fig. 5.3.
Criteria for the occurrence of
various sediment transporta-
tions. A: Threshold to the indi-
vidual particle motion, B: Thres-
hold for the sediment gravity
flow, C: Threshold for the occur-
rence of debris flow, D: Threshold
by Tognacca et al. (2000), E: Thres-
hold by Arumanini et al. (2003)
Chapter 5 Progress in Debris Flow Modeling 63

found on the boundary between the deposits A and B,

5.4 Processes in Transformation and Stoppage
and after the stoppage of the earth block the debris flow
of Landslide-induced Debris Flow
passed and then stopped on the already halted earth block.
Some parts of debris flow overflowed the dam and caused
A severe rainfall from 6th to 10th July, 1997 amounting the severe disasters. When landslide occurred, rainfall had
about 400 mm attacked Southern Kyushu, Japan. On already ceased about 4 hours before, and hence, the river
10th July, 4 hours after the stopping of rainfall, a large land- discharge had to be little. It must not be enough to gener-
slide of 80 m in the maximum width, 190 m in length, and ate debris flow by the erosion of halted earth block. There-
30 m in the maximum depth occurred in the Harihara fore, the debris flow should have been generated from the
River basin. There was a dam to check debris flows, but sliding earth block without water supply from outside.
the volume of the landslide (130 000 m3) was so large that The description above suggests a model for the gen-
some parts that transformed into debris flow overflowed eration of landslide-induced debris flow as illustrated in
the dam. The debris flow destroyed 19 houses and killed Fig. 5.7. If a landslide is triggered under a plenty of
21 people (Fig. 5.5). After the disaster a trench cut inspec- groundwater within the slid earth block, liquefaction
tion of the deposit upstream of the dam was carried out. may take place near the slip surface; it may easily occur
It revealed that there were two different deposit units as if the earth block has rich void space. Velocity distribu-
shown in Fig. 5.6; one was the relatively undisturbed earth tion in the liquefied layer must be as shown in the upper
block in the lower part (deposit A) and the other was the diagram; zero at the bottom and the maximum at the
disturbed debris flow deposit (deposit B) laid on the boundary between the liquefied layer and the earth block,
deposit A (Sabo Technical Center 1998). This fact suggests so that the earth block supported by the pressure within
that the considerable parts of solid earth block reached the liquefied layer goes faster than the liquefied layer
the trench cut position with little disturbance; the small and the liquefied layer is left behind as debris flow. The
trees originally on the surface of mountain slope were motion of earth block continues as long as the slope

Fig. 5.5.
The Harihara River debris flow
64 Tamotsu Takahashi

Fig. 5.6.
Sketch of the inside structure of
deposit at the trench wall

Fig. 5.7. Process model of landslide earth mass transformation into

debris flow

gradient is steep and liquefaction at the boundary be- Fig. 5.8. A moving earth mass on a slope with the partial liquefaction
tween the earth block and the ground continues. Thus,
the earth block becomes thinner and thinner with its between the particles must be filled with water or slurry.
advance down the slope cannibalizing itself to left de- The condition of no water supply from the outside of the
bris flow behind. When the earth block arrives at a flat- layer to be liquefied requires that the skeletal structure of
ter area, its motion stops, and the succeeding debris flow the mass must be full of pores so as to be able to form the
gets over it and continues to run down. Therefore, the mixture of well dispersed particles and fluid. If the mass
earth block remains underneath of the debris flow de- is not saturated by water, the liquefaction must occur with
posit. In the Harihara River debris flow, the earth block volume contraction.
was worn out from its maximum thickness of 30 m to The equation of conservation of momentum for the
about 5 m during the motion of about 200 m and stopped earth block is as following:
upstream of the sabo dam.

5.5 Simplified Mathematical Model

For the sake of simplicity, a moving earth mass of rectan- (5.14)

gular parallelepiped is considered on a uniformly inclined
plane as shown in Fig. 5.8. The lowest part of the earth where Ue is the velocity of the earth mass, hl the thickness
mass is liquefied by the shearing action between the mass of liquefied layer, ul the mean velocity of liquefied layer,
and the slip surface. To accomplish the liquefaction, the H the thickness of the mass, L the length of the mass, hs
skeletal structure of the mass must be destroyed and the the thickness of the saturated layer in the mass, ha the
particles must become free from others. The void space thickness of debris flow that follows the moving mass, ua
Chapter 5 Progress in Debris Flow Modeling 65

the mean velocity of debris flow, t the apparent density (5.18)

of the mass, l the apparent density of the liquefied layer,
a the apparent density of debris flow, k the kinetic fric-
tion coefficient of the mass, f the resistance coefficient of
the liquefied layer, a coefficient, and Fside the friction at
the side wall. In Eq. 5.14, the left-hand side term repre- where is the viscosity of the liquefied layer. Then,
sents the change in the momentum of the mass. The first
term of the right-hand side is the driving force due to
gravity operating on the mass. The second term is the
friction force at the bottom due to the load that is directly
transmitted through the particles. The third term is the As a field standard, if one assumes c=3510 5 Pa,
momentum that is left behind. The fourth term is the fluid =100 Pas, hl=100 cm and ul=1000 cms1, one obtains
resistance of the liquefied layer. The fifth term is the mo- =0.03.
mentum supply from the following debris flow (this term This simplified one-dimensional model was extended
becomes zero if ua Ue). The sixth term is the hydrostatic to analyze the motion across a three-dimensional ter-
pressure operating to the mass, and the seventh term is rain, in which we considered a rigid body is moving on
the side friction. the surface of a fluid flow interacting each other and the
As an extreme case, if all the load of the mass is trans- motion of fluid was analyzed by the Eulerian continuous
mitted to the bottom as the fluid pressure (perfectly liq- fluid equations and the motion of the earth mass was
uefied case), becomes zero. If the mass is saturated but analyzed by the Lagrangian solid mass equations (Taka-
no liquefied layer is produced, = ( )C* /{( )C* + }, hashi et al. 2003). This hybrid simulation model was veri-
and if the mass is not saturated by water even at the bot- fied by the laboratory experiments and it was success-
tom layer, becomes 1. In an actual case would have a fully applied to reproduce the actual debris flow occurred
value between one and zero; it would depend on the par- at Hogawati, Minamata City, Japan on 20th July 2003 (Ta-
ticle concentration and the diameter. The smaller the , kahashi 2004).
the milder the slope the mass can reach. If is small and
hs is nearly equal to H, almost all of the mass will be trans-
formed into debris flow. 5.6 Models for Debris Flow Dynamics
The conservation equation of the mass volume is given as
Aforementioned discussions excluded the details of re-
sistance coefficient to flow that will depend on the char-
acteristics of flow such as the size and distribution of com-
posing material, solids concentration, velocity, etc. Here-
and the change of the thickness of liquefied layer is given as after, the existing models of debris flow are reviewed cat-
egorizing into single-phase flow model and the two-phase
flow model and applying a two-phase model that consid-
ers the debris flow is comprised of the mixture of two
continuum media of fluid phase and solid phase debris
where i is the velocity at which the mass is abraded at the flows are classified and made it possible to analyze re-
boundary between the bottom of the mass and the upper spective types of debris flows.
surface of the liquefied layer. Many simulation models for granular flow with the
The shearing stress from the liquefied layer operates negligible effects of interstitial fluid like that in vacuum
to the bottom of the mass, and this shearing stress, , or heavy particles motion in the atmosphere have been
does the work ul in unit time. Because the mass has a given, that is to say single-phase model. They can be cat-
certain strength, s, the amount of energy of is must be egorized into the continuum media model and the dis-
applied to the mass to be abraded with the rate i. There- crete particle model. The former describes the granular
fore, i would be regulated by the relation is=ul. If one assembly as a continuum media and its characteristics
writes =/s, the following relation is obtained: in flow are analyzed by the Eulerian forms of the conti-
nuity and the momentum conservation equations. The
(5.17) characteristics of discrete particles motion are neces-
sarily neglected. The latter one, on the other hand, traces
If the strength of the mass is determined by the cohe- the motion of individual particles by the Lagrangian
sion, c, and if the liquefied layer is considered as a equations.
Newtonian fluid, the following relationships will be sat- In the continuum media model, the characteristics of
isfied: flow as a granular assembly are indirectly involved in the
66 Tamotsu Takahashi

stress terms in the Eulerian momentum equation. These model becomes Euler-Euler coupling model. In this model,
terms are obtained from the relationship between the two momentum conservation equations each for liquid and
stress and the strain of the continuum that is called the solid phases are given as following (Iverson 1997):
constitutive equation. Although the general constitutive
equation under the arbitrary conditions of particle den- (5.21)
sity, particle size distribution, and particle interactions is
difficult to obtain, if an appropriate form under a par-
ticular flow condition is given, the continuum media (5.22)
model can give an operative method to explain the flow
characteristics en masse under that condition. where vs is the velocity of solids, vf the velocity of fluid, g
The discrete particle model describes the motions of the acceleration due to gravity, Ts the stress tensor of solid
individual particles by the Newtons second law equa- phase, Tf the stress tensor of fluid phase, and F the in-
tion under the effects of contact forces among the neigh- teraction force per unit volume that results from the mo-
boring particles. Two kinds of approach for the inter- mentum exchange between the solid and the fluid con-
particle contact have been adopted: the hard-contact stituents.
technique and the soft-contact technique. In the hard- Addition of these two equations yields a momentum
contact technique, particles behave as rigid body and conservation equation applicable to the bulk mixture. This
the contacts arise only between the two adjacent parti- procedure eliminates the assessment of complicated in-
cles in an instantaneous moment (Campbell and Brennen teractions between fluid and solid phases. If nearly steady
1985; Straub 2001). This technique has a shortcoming uniform flow is considered and the relative velocity be-
that it cannot handle the simultaneous collision of mul- tween solids and fluid is neglected in Cartesian coordi-
tiple bodies. nate system, following two equations are finally obtained
The soft-contact technique is free from the defect of (Takahashi 1991):
hard-contact technique (Campbell 2001). At the moment
of soft-contact, particles deform a little and the forces (5.23)
proportional to the stiffness of particles act on the con-
tact points. The interaction between particles on a (5.24)
contact point is modeled by the deformation of the in-
terface instrument that consists of a spring and a dash- where Pds is the pressure in solid phase, Pf the pressure in
pot connected in parallel, and a pair of such an instru- fluid phase in excess over hydro-static one, Tds the shear
ment is mounted normally and tangentially on each stress in solid phase, Tf the shear stress in fluid phase, T
contact point. To handle non-cohesive particle, a joint {= ( )C + } the apparent density of debris flow ma-
that is nonresistant to tension toward normal direction terial. These equations are the fundamental equations in
and a slipping joint that is effective under a tangential the two-phase fluid model describing the stress balance
force more than a threshold should be serially connected in the flow of mixture. The essential problem in this two-
to the respective instruments (Gotoh and Sakai 1997). phase mixture theory is how the left-hand sides of Eqs. 5.23
Thus, many parameter values must be given beforehand, and 5.24 are described.
and if particle number becomes large, load on computer If solid phase is considered as an assembly of discrete
becomes enormous. However, the implementation of particles, flow should be analyzed by Euler-Lagrange cou-
simulation makes possible to observe the motion of pling model, in which the individual particles motion is
individual particles that may be difficult to observe in solved by the discrete element method mentioned earlier
the physical model tests. The quantitative characteris- taking the interaction between particles and fluid into
tics of overall flow of bulk mixture such as depth and account. Some investigations on the bed load transport
mean velocity can be obtained only after the completion have been done by this method (Gotoh et al. 1994), but
of simulation. debris flow has not been analyzed by this method.
In the case of debris flow, the interstitial fluid is water
or slurry, so the relative density of particles in flow is small,
and therefore, the effects of interstitial fluid often play 5.7 Single-phase Continuum Models
important role. Then, flow should be considered as that
consists of two phases. Liquid phase is naturally described In the single-phase continuum models, the mixture of
by the Eulerian equation, but as mentioned above, the solid particles and fluid is considered as a kind of continuum
phase may be described as a continuum media or as an fluid whose constitutive relation between stress and rate
assembly of discrete particles. of strain is to be found empirically by the laboratory tests
If solid phase is considered as a continuum media and of samples collected from flow or by the deduction from
described by the Eulerian equations, mathematically the the back analysis of the model applied to the flow in field.
Chapter 5 Progress in Debris Flow Modeling 67

5.7.1 Visco-plastic Fluid Model

Debris flow that induced by landslide initiates from the

gradual or the abrupt motion of a debris block under the
effects of water, and after a motion of considerable dis-
tance, it comes out to flat area and stops. Therefore, if one
considers the material has a strength intrinsically, one can
explain the starting of motion attributing to the opera-
tion of a stress more than the (yielding) strength, and flow
stops if the stress in flow becomes smaller than the strength.
Debris flow models that consider the material is com-
prised of a kind of viscous fluid having strength are called
visco-plastic fluid model, and they assume the relation
between the stress and the strain rate as following:

Fig. 5.9. Velocity distribution in Bingham fluid flow
where y is the yielding strength, u the velocity at height z.
K and n are the numerical coefficients.
The simplest in the visco-plastic models is the one set- 5.7.2 Dilatant Fluid Model
ting n = 1, that is called Bingham fluid model. This model
was first proposed independently by Yano and Daido Debris flow that consists of mainly coarse grains larger
(1965) and Johnson (1965). Later, Johnson (1970) modi- than gravel size can often reach to an area flatter than
fied his model to the following Coulomb-viscous model: about 4 keeping its high mobility and competence to
freight big boulders. Takahashi (1978) inferred such small
(5.26) resistance to flow was caused by the dispersion of grains
separated with one another, and the inter-particle colli-
where n the internal normal stress and B the Bingham sion gave rise to the dispersion of heavier grains than the
viscosity. ambient fluid.
Visco-plastic fluid models, including the Herschel- Bagnold (1954) was the first to discover the significance
Bulkley fluid model that sets n < 1 (Coussot 1995) and of inter-particle collision as the cause of particle disper-
the polynomial models that add the term (du/dz)2 to Bing- sion. His experiments and a simple binary grain collision
ham fluid model (Julien and OBrien 1997) or add (du/dz)n analysis gave the stress-strain relationships in inertial and
to Eq. 5.26 (Chen 1988), are still frequently used especially viscous regimes. In the inertial regime, those relationships
for the cases of viscous type debris flow. were given as following:
For example, the stress balance equation for steady
uniform Bingham fluid flow is given by (5.28)

(5.27) (5.29)

The integration of Eq. 5.27 under the boundary con- where p is the dispersive pressure in flow, ai the numeri-
dition; u = 0 at z = 0, gives the velocity distribution as cal coefficient, i the collision angle, the linear concen-
shown in Fig. 5.9, in which z' (y /T gsin) is the thick- tration of grains defined by = {(C* /C)1/3 1}1. He di-
ness of the plug that behaves like a rigid body. rectly measured the pressure within the flow suspending
The serious shortcoming of visco-pastic models is the neutrally buoyant grains that was in excess of the pres-
difficulty in determining the crucial parameter values sure in plain fluid flow, so that p in Eq. 5.28 is equivalent
such as y, B, etc. Those values obtained by the labora- to Pds in Eq. 5.23. He could not directly measure Tds cor-
tory tests and by the field observations are usually much responding to Pds, but it was obtained by subtracting the
different, and they cannot account for the changes of pa- periphery drug in the plain fluid flow from the periphery
rameter values due to addition or subtraction of water or drug in the mixture flow. Therefore, thus deduced value
sediment during motion. According to the authors ob- in Eq. 5.29 is equivalent to Tds in Eq. 5.24.
servation, the typical viscous type debris flow in the Bagnold in his experiment used neutrally buoyant solid
Jiangjia ravine, China lacks the plug that should exist in a spheres immersed in a Newtonian fluid, but Takahashi
visco-plastic fluid flow (Takahashi 1999). (1978) exploited his experimental results to the open chan-
68 Tamotsu Takahashi

nel flow in which heavy particles are very densely con-

centrated, and he also assumed the uniform distribution
of grains in the entire depth. Then, the stress balance
equations for the normal and the parallel directions to
the bed were obtained, respectively, as following:



Referring to the discussion above, this model is an

equivalent model to the two-phase model in which the
dynamic fluid effects; Pf and Tf, are neglected. A kind of
fluid having the relationship; ~ (du/dz)n, n > 1, is called
dilatant fluid, so Takahashis model is called the dilatant
fluid model.
Equations 5.30 and 5.31 are integrated under the
boundary condition; u = 0 at z = 0, to obtain the flow ve- Fig. 5.10. Velocity distribution of debris flow nearly in equilibrium
locity at height z. The result of integration of Eq. 5.31 is concentration on rigid bed
given as
in the equilibrium stage, the dynamic collision angle is
almost equal to the static internal friction angle. There-
fore, one can consider Eq. 5.13 is the equilibrium grain
concentration formula obtained from the static as well as
Mathematically, the velocity distribution function ob- the dynamic equilibrium conditions. Thus, the dilatant
tained from Eq. 5.30 has not the same expression with fluid model can well explain the dynamics of debris flow
the one obtained from Eq. 5.31. These two functions be- in which inter-particle collision plays the dominant role,
come physically equivalent only when the following for- moreover, different from visco-plastic fluid model, it can
mula is satisfied: account for the change in solid concentration during
motion (Takahashi 1991).
The assumption of uniform grain distribution through-
out the depth is, however, at least theoretically, too much
simplification, and this causes the contradiction that there
For example, if one substitutes = 2.65 g cm 3 , exists two independent formulae 5.30 and 5.31 for one
= 1.0 g cm3, tani = 0.75 and = 18 into Eq. 5.33, one unknown velocity (Iverson and Denlinger 1987). To dis-
obtains C = 0.46. Then, if one supplies the mixture of water solve this contradiction, Takahashi (1991) assumed the
and grains whose grain diameter is 4 mm and the vol- following formula:
ume concentration is 0.45 into a rigid bed flume of 18 in
gradient, one observes velocity distribution that fits very (5.34)
well to Eq. 5.32 by giving ai = 0.04 that was obtained by
Bagnold for wax beads as shown in Fig. 5.10. As another This formula means the dynamic collision angle be-
example, if one supplies water with a steady rate onto an tween particles is a little larger than the static internal
erodible bed that is saturated by water and having the friction angle when C is smaller than C*. This would be a
slope gradient steeper than 15, debris flow develops reasonable assumption because the wider space between
downstream and approaches an equilibrium state as il- particles than in the static bed will result in the deeper
lustrated in Fig. 5.1. As explained earlier, in the steady colliding angle.
equilibrium flow region grain concentration is given by Substitution of Eq. 5.34 into Eqs. 5.28 and 5.29 results
Eq. 5.13 and the validity of Eq. 5.13 as the equilibrium in rather good agreement with Bagnolds experiments,
grain concentration formula has been proved by many and also the substitution of Eq. 5.34 into Eqs. 5.30 and
flume experiments. This fact and the good fitness of 5.31 gives two independent formulae necessary and suf-
Eq. 5.32 under Eq. 5.33 to the rigid bed experiments sug- ficient to obtain u and C at height z. The analysis using
gests that in the equilibrium stage tani = tan. Hence, these two equations can well explain the vertical grain
Chapter 5 Progress in Debris Flow Modeling 69

5.8.1 Coulomb Mixture Theory (Quasi-static Debris Flow)

Iverson and Denlinger (2001) claims that the Coulomb

friction term; s, is almost always larger than the particle
collision term; c, stating that the Savage number, defined
by Eq. 5.36, in various field examples and the large scale
experiments by USGS is smaller than 0.1 as the evidence,
and they develop a Coulomb mixture theory as the only
one suitable model for debris flow.


Fig. 5.11. Theoretical velocity distribution and the experimental data The numerator of Eq. 5.36 is, as is clear from Eq. 5.28,
the term representing the particle collision effects, and
concentration and velocity distributions on an erodible the denominator represents the load operating on the bed
bed. The velocity distribution formula is given by due to submerged weight of particles. They say the de-
nominator is the gravitational grain contact stress that
produces Coulomb friction, but it must be noted that the
static gravitational contact stress can be transferred from
the surface to the bottom of flow only when a skeletal
where us is the velocity at the flow surface and Z = z/h. structure due to the enduring contact of grains extends
Equation 5.35 is compared with the experimental data on throughout the depth. To furnish this necessity, the local
erodible bed in Fig. 5.11. The vertical velocity distribu- solid fractions everywhere within flow must exceed a
tion on erodible bed has a peculiar form with two inflec- threshold level C3. According to Bagnold (1966), C3 for
tion points; one near the bed and the other near the sur- natural beach sand is about 0.51. It is true that denomi-
face of flow, whereas that on rigid bed, especially when nator represents the effective pressure at the bottom of
flow contains slightly thinner concentration than equi- flow, but it is not necessarily transferred as the static gravi-
librium, has simple concave form (Fig. 5.10) as that is ex- tational contact stress. It is transferred as the collision
plained by the first approximation under the assumption stress if grain concentration is smaller than C3. If Ccos
of uniform concentration distribution. is multiplied to the denominator, the term precisely rep-
resents the effective pressure at the bottom, but the term
multiplied by Ccos to the numerator does not represent
5.8 Two-phase Fluid Flow Model (Mixture Theory) the magnitude of collision stress. To represent the colli-
sion stress not Ccos but ai cosi 2 must be multiplied
As mentioned earlier, how one considers the left-hand side to the numerator as is evident in Eq. 5.28. As the conse-
terms in Eqs. 5.23 and 5.24 is the origin of diverse debris quence, Eq. 5.30 indicates the effective pressure is bal-
flow theories. The particle dispersive pressure due to in- anced with collision stress. Therefore, Ns that is smaller
ter-particle collision; pc, and the static skeletal pressure than 0.1 cannot be the proof of the predomination of
due to enduring contact between particles; ps, would be Coulomb stress term. Coulomb mixture theory is con-
the candidates to consist Pds. Pf would exist if the sedi- sidered as a model suitable for very densely grain con-
ment mass contracts volumetrically or if there is a net centrated quasi-static flow.
flux of sediment toward the bed (Iverson 1997). Herein, Herein, we do not dwell on the details of Coulomb
only quasi-steady flow is considered, so Pf = 0. The shear- mixture theory.
ing stress due to inter-particle collision; c, the shearing
stress due to enduring contact between particles; s, and
the streaming stress due to migration of particles in one 5.8.2 Classification of Dynamic Debris Flows
layer to other layer; k, would be the candidate to consist
Tds. The deformation stress within interstitial fluid; , Dynamic debris flow herein defined is the one in which
and the turbulent mixing stress; t, would be the candi- dynamic stresses predominate in governing the behav-
date to consist Tf. All such candidates cannot prevail si- iors of flow. The total shear stress (= Tds + Tf) in a grain
multaneously, but some among others prevail in a par- and water mixture would be described as
ticular regime. Therefore, via the discussion of dominant
stresses, debris flows can be classified. (5.37)
70 Tamotsu Takahashi

If C is larger than about 0.2, k becomes smaller than where Ba (= 1/2dp2(du/dz)/) is called Bagnold number
c and the difference between the two tends to become and Re (= TUh/) is Reynolds number, in which l h and
drastically large as C becomes larger (Takahashi and l(du/dz) U are assumed.
Tsujimoto 1997), and if C is smaller than about 0.5, s is Summarizing the above discussions, one can conclude
almost equal to zero (Bagnold 1966). Therefore, in a dy- that dynamic debris flows occur in the domain of the ter-
namic debris flow, if the average grain concentration (ex- nary diagram as shown in Fig. 5.12, where the three axes
cept for the fine constituent making up the interstitial represent relative depth, Bagnold number and Reynolds
slurry) is between 0.2 and 0.5, and if t /c and /c are number, respectively. The region of large Bagnold num-
small, c / 1 is satisfied. Debris flow in which the stress ber and small relative depth is that for stony debris flow.
due to inter-particle collision dominates among others is The region of small Bagnold and Reynolds numbers is
called stony debris flow. On the other hand, if t / 1 is that for viscous debris flow, and the region of large
satisfied, it is called turbulent muddy debris flow, and if Reynolds number and large relative depth is that for tur-
/ 1 is satisfied, it is called viscous debris flow. bulent muddy debris flow. Thus, the areas close to the
Equation 5.29 and the other previous theoretical con- three apices are the regions of stony, viscous and turbu-
siderations on the constitutive relations of inertial granu- lent muddy debris flows, respectively, and the other area
lar flows confirm (Campbell 1990) in the triangle is the region of hybrid debris flows.
Previous experiments reveal that if the transport con-
(5.38) centration of large grains is smaller than about 0.2 (herein,
large grains are defined as the ones not being suspended by
and fluid mechanics give the following forms: turbulence, and the transport concentration is defined as
the ratio of large grains discharge to the total discharge of
(5.39) large grains plus fluid phase including suspended sediment),
large grains cannot disperse in the entire depth of flow. The
(5.40) mixture layer of large grains and fluid appears only in the
lower area of flow. This type flow is, as mentioned earlier,
where f1(C) and f2(C) are the functions of C, the vis- immature debris flow in contrast to (mature) debris flows
cosity of interstitial fluid, and l the turbulent mixing in which large grains are dispersed in the entire depth.
length. From these equations one obtains the following Thus, sediment transport or sediment moving phe-
formulae: nomena can be categorized by the transport concentra-
tions into the plain liquid flow, the individual particle
motion (bed load and suspended load; 0 < C 0.02), the
immature debris flow (0.02 < C 0.2), the debris flow
(0.2 < C 0.5), the quasi-static debris flow (0.5 < C 0.6),
and the sliding of rigid earth body (0.6 < C). The bound-
ary C values indicated are only for reference, they must
depend on properties of particles such as cohesion, shape
and size distribution.

5.8.3 Generalized Theory for Inertial Debris Flow

Debris flows that occur in the domain adjacent to the axis

representing the relative depth in Fig. 5.12 can be generi-
cally called inertial debris flow, because the inertial
terms; c or t, dominate in flow. The hybrid flow in the
inertial debris flow consists of the lower particle colli-
sion layer and the upper turbulent suspension layer as
illustrated in Fig. 5.13. The ratio of these two layers in the
depth depends on the relative depth; h/dp, and the con-
centration. If the relative depth is small and the particle
collision layer occupies almost entire depth, it is stony
debris flow, and if the relative depth is large and the tur-
bulent suspension layer occupies almost entire depth, it
is turbulent muddy flow. Herein, the theory of Takahashi
Fig. 5.12. The domains of existence of various types of debris flows and Satofuka (2002) is introduced.
72 Tamotsu Takahashi

If calculated solids concentration is larger than Clim in mental results. In the experiments debris flows were gen-
the entire depth, even the criterion 1) is satisfied, there is erated by abruptly supplying water from upstream onto
no upper layer, and h2 = h. If, at a certain height, calcu- the water saturated erodible bed in a flume of 9.9 cm wide
lated solids concentration becomes smaller than Clim, but and 10 m long. Bed slope was changed, and the discharge
the criterion 1) is not satisfied, it is immature debris flow. of sediment plus water; QT, was held constant at about
Theoretical solids concentration and velocity distri- 531 cm3 s1. Four kinds of materials; 0.201, 0.066, 0.030
butions were obtained corresponding to the conditions; and 0.017 cm in median diameters with densities between
= 2.65 g cm3, = 1.0 g cm3, tan = 0.7, e = 0.3, = 0.3, 2.64 and 2.66 g cm3 and internal friction angles between
C* = 0.65, and C3 = 0.5. Figure 5.14 shows the examples of 38.5 and 39, were used for the erodible bed. The debris
the calculated results, where flow depths are held constant flow samples were collected with a bucket at the outlet of
at 2 cm throughout the cases. When particle diameter is the flume to measure QT and Ctr. As shown in the figure,
7 mm, even for steep channel slope of 18, the condition the theory can well explain the tendency that the smaller
for particle suspension is not satisfied but particles are the particle diameter, the larger the equilibrium trans-
dispersed in the entire depth due to the action of inter- port concentration becomes under a given channel slope.
particle collision, and the velocity is small. It is stony de- In the same figure, other than those corresponding to
bris flow. For the same size particles, when channel slope the experiments, the theoretical curves for dp = 1 cm and
is 6, particles cannot be dispersed in the entire depth, 0.5 cm are also drawn. Under the experimental conditions
and it is immature debris flow. When particle diameter is herein, particles larger than 1 cm in diameter cannot be
1 mm, even the channel slope is as flat as 6, the condition suspended by turbulence even on the steep slope of 15.
for particle suspension is satisfied, and there exists the The theoretical curve for dp = 1 cm almost coincides with
upper layer in which particles are in suspension. It is hy- the respective concentrations calculated by substituting
brid type debris flow. 0.8 into tan in Eq. 5.13 or into tani in Eq. 5.33 and into
Figure 5.15 compares the theoretically obtained veloc- tan in the following equilibrium concentration formula
ity distributions with the experimental data of Hirano for immature debris flow (Takahashi 1991):
et al. (1992), where the velocity is normalized by the sur-
face velocity. Debris flows in the experiments were gen- (5.52)
erated by supplying water from upstream onto the water
saturated erodible bed flume 20 cm in width and 7 m in where C is given by Eq. 5.13 and it is valid only when Ctr
length. Bed slope was held constant at 14. The relative by this equation is smaller than C calculated by Eq. 5.13.
depth was changed by changing both the supplying wa- The reason for branching of curves corresponding to
ter discharge and the particle size. The theoretical as well dp = 0.5 cm and 0.2 cm from the curve for dp = 1 cm is due
as experimental velocity distribution curves in the figure to the beginning of suspension on the slopes steeper than
have a tendency to become lower the break point on the those corresponding to the respective branching points.
curve; i.e., the boundary between the upper and the lower
layer, and to increase the degree of concavity as the rela-
tive depth increases. Moreover, the theoretical curves well
fit to the experimental results.
Figure 5.16 is a comparison of the theoretical equilib-
rium transport concentrations; Ctr, with authors experi-

Fig. 5.14. Velocity and particle concentration distributions by the Fig. 5.15. Theoretical and experimental velocity distributions on
theory erodible bed
74 Tamotsu Takahashi

to particle flux; Nc, perpendicular to the main flow direc- Substitution of Eq. 5.63 into Eq. 5.62 yields the par-
tion as following: ticle concentration distribution function as following:

where = du/dz, and Kc the numerical coefficient. The
word encounter in this explanation does not necessarily where
mean the collision, but only an approach of particles can
give rise to squeezing flow in the interstitial fluid that
moves the particle perpendicular to the main flow.
If coarse particles distribute anisotropically, the ap- (5.65)
parent viscosity varies spatially. Because the apparent vis-
cosity is larger in the higher concentration region than in
the lower concentration region, the resistance to particle
migration into the higher concentration region is larger Figure 5.17 shows the examples of solids concentra-
than into the thinner concentration region. This mecha- tion distribution calculated under the two combinations
nism gives rise to particle flux; Na, as following: of parameters; C* = 0.6, Cb = 0.57, = 1.38 g cm3 ( 0.9),
and C* = 0.72, Cb = 0.70, = 0.77, respectively. The respec-
tive combinations of parameters correspond to the con-
ditions in our laboratory experiment and in the Jiangjia
Gully, China. For both cases, Kc = 0.5 and Ka = 0.75 are
where Ka is a numerical coefficient. adopted from the experimental results obtained by
The coarse particles tend to deposit due to gravity, and Phillips et al. (1992). The concentration at the bottom of
the flux is given as following: flow; Cb, is assumed a little smaller value than C*, because
the substitution of C = C* into Eq. 5.64 cannot make sense.
As is clear in Fig. 5.17, when the channel slope is steep,
very high solids concentration maintains up to the sur-
face of flow, and for relatively mild slope channel, the high
where G(C) is the hindrance function to account for highly concentration abruptly decreases in the upper region. This
concentrated group settling, and here is the case of immature viscous debris flow.
The coarse particle concentration in a given slope
(5.60) channel becomes large and uniform as becomes small.
Namely, the denser the interstitial fluid, the flatter the area
is assumed, where is the specific viscosity that is given of debris flow of a given solids concentration will reach.
by (Krieger 1972) The equilibrium coarse particle concentration for a
given channel slope can be obtained by integrating
Eq. 5.64. The results corresponding to the conditions used
in obtaining Fig. 5.17 are shown in Fig. 5.18. The respec-
tive curves show the solids concentration can easily be
In a steady state, these three fluxes must balance so as saturated if channel gradient becomes steeper than a cer-
to perform zero particle flux perpendicular to the main tain value. The value of saturation is, as is clear in Fig. 5.17,
flow direction. This condition is given by approximately equal to Cb. This fact suggests that, as long


The balance of force equation in such a steady uni-

form flow is given by


Fig. 5.17. Distributions of coarse particle concentrations on vari-

where = ( )/. ous slopes
Chapter 5 Progress in Debris Flow Modeling 75

as flow rate is sufficiently large, once a saturated flow is form from bottom to a height (Z = Z2), it has the tendency
formed in an upstream steep channel reach, it runs down like shown by the line in Fig. 5.19. It is given by the fol-
the gradually flattening channel reach keeping its original lowing equation:
concentration until the slope becomes flatter than a critical
one on which the equilibrium concentration is smaller than
the saturated value. Deposition commences if the solids
concentration is larger than the equilibrium value.
The velocity distribution can be obtained by solving
Eq. 5.63 under the boundary condition; at Z = 0, u' = 0,
where u' = u/u*. The solution cannot be obtained analyti-
cally, but provided the viscosity is approximated as uni- (5.66)

where X = (1 + C /a) 1 and C is the mean particle
concentration below Z2.
Well mixed debris flow material in the Jiangjia Gully
except for the component larger than 10 mm was imitated
by mixing silica sand and kaolin. This material was mixed
with water to make the constant solids concentration of
0.56~0.57, and it was supplied into the upstream end of
experimental flume from a hopper.
The velocity distributions given by Eq. 5.66 are
compared with the experimental data in Fig. 5.20, in
Fig. 5.18. Equilibrium coarse particle concentrations versus chan- which Z2 = 1 and a = 8.1 Pa s obtained from the theory
nel slopes explained above are used. Figure 5.20a is the case the
flow is produced on the deposit formed by a previous
debris flow surge, and Fig. 5.20b is the case the flow is
produced on rigid bed. This figure shows that not only
the velocity distribution obeys the Newtonian laminar
flow equation equally well on erodible and rigid beds,
but also the absolute velocity values are well predicted
by this theory.
Recently, Armanini et al. (2003) did the similar theo-
retical and experimental discussion that took the yield
strength of interstitial fluid into account.

5.9 Conclusion

The processes of initiation and the mechanics of flow of

Fig. 5.19. Schematic velocity distribution in an immature viscous various types of debris flow were discussed from the point
debris flow of view of systematizing the theory of debris flows.

Fig. 5.20.
Calculated and experimental
velocity distributions
76 Tamotsu Takahashi

The causes of initiation of debris flows were classified depends on both the channel slope and the particle di-
into the erosion type and the landslide-induced type. The ameter. The theoretically obtained results rather well co-
process of formation of the erosion type debris flow can incide to the experimental results.
be analyzed similar to the known methods for the analy- The viscous type debris flow was modeled as a lami-
ses of fluvial processes by the bed load transportation. nar flow of a Newtonian fluid, in which coarse particles
The differences between the analyses of debris flow and were dispersed in flow almost uniformly under the bal-
bed load transport appear in the flow resistance law and ance between the upward dispersing fluxes due to squeez-
the erosion velocity equation. ing flow and viscosity gradient on the encounter of par-
A model of liquefaction of the landslide earth mass ticles embedded in the two adjacent shearing layers and
was introduced. At least the lower part of the slid earth the downward flux due to gravity settling. The theory not
mass is assumed saturated with water. The saturated part only well explains the high fluidity of very densely par-
will liquefy from lower to upper by the effect of shearing ticle freighting debris flow but also predicts the velocity
between the mass and the ground. The liquefied layer not of viscous debris flow without using the empirical param-
only acts as a lubricant to move the mass on top of it, but eters by contrast to many previous visco-plastic models.
also gets behind the mass to form a following debris flow.
Therefore, if the mass is completely saturated and lique-
faction proceeds to the extent of disappearance of the References
mass, the mass completely transforms into debris flow. If
the liquefaction ceases halfway, the following debris flow Armanini A, Gregoretti C (2000) Triggering of debris-flow by over-
land flow; A comparison between theoretical and experimental
may overflow the stopped mass. The liquefaction without
results. In: Wieczorek GF, Naeser ND (eds) Debris flow hazards
supply of water from the outside of the slid earth mass will mitigation: mechanics, prediction, and assessment; Proceedings
be accomplished under the contractive destruction of the International Conference, Taipei, Balkema, Rotterdam, pp 117124
skeleton of the mass. This rationale leads to an inference Armanini A, Dalri C, Fraccarolo L, Larcher M, Zorin E (2003) Ex-
that the mass of dilative nature will not transform into de- perimental analysis of the general features of uniform mud-flow.
bris flow without supply of water from the outside. In: Rickenmann D, Chen CL (eds) Debris flow hazards mitiga-
tion: mechanics, prediction, and assessment. Proceedings Inter-
The basic system of equations to describe the afore- national Conference, Davos, Millpress, Rotterdam, pp 423434
mentioned process was given. Although the detailed com- Ashida K, Takahashi T, Mizuyama T (1977) Study on the initiation of
parison of the theory to the laboratory experiments were motion of sand mixtures in a steep slope channel. J JSECE 29(4):
abbreviated due to limitation of pages, the validation of 613 (in Japanese)
the theory including its expanded form capable to ana- Bagnold RA (1954) Experiments on a gravity-free dispersion of large
solid spheres in a Newtonian fluid under shear. Proc R Soc Lon-
lyze the three-dimensional motion on a topographically don, Ser A 225:4963
complicated ground is referred by introducing authors Bagnold RA (1957) The flow of cohesionless grains in fluid. Philos
recent investigations. Trans R Soc London, Ser A 249:235298
Various concepts for modeling debris flows are re- Bagnold RA (1966) The shearing and dilatation of dry sand and sing-
viewed. Under the present state of the art, the mixture ing mechanism. Proc R Soc London, Ser A 295:219232
Campbell CS (1990) Rapid granular flows. Annu Rev Fluid Mech
theory that considers the debris flow material is a mix-
ture of two continua; one is the interstitial fluid and the Campbell CS (2001) Granular flows in the elastic limit. Spec Publs
other is a kind of continuum media comprised of granu- Int Ass Sediment 31:8389
lar assembly, is influential. By the adoption of appropri- Campbell CS, Brennen CE (1985) Computer simulation of granular
ate constitutive relations in the mixture theory, debris shear flows. J Fluid Mech 151:167188
flows were classified into three; stony type, turbulent Chen CL (1988) Generalized viscoplastic debris flow. J Hydraul Eng-
ASCE 114(3):237258
muddy type and viscous type, and their respective exist- Coussot P (1995) Structural similarity and transition from New-
ing regions were defined in the ternary diagram whose tonian to non-Newtonian behavior for clay-water suspensions.
three axes represent the Bagnold number, the Reynolds Phys Rev Lett 74:39713974
number, and the relative depth. Gotoh H, Sakai T (1997) Numerical simulation of sheetflow as granu-
The inertial debris flow was defined as the unified con- lar materials. J Waterw Port C Div 123(6):329336
Gotoh H, Tsujimoto T, Nakagawa H (1994) Numerical model of in-
cept of the stony and the turbulent muddy debris flows. terface momentum transfer and interparticle collision in bed-
The theoretical solutions of the inertial debris flow; the load layer. In: Proceedings APD-IAHR, pp 565572
particle concentration distribution, the velocity distribu- Hirano M, Hashimoto H, Fukutomi K, Taguma K, Pallu MS (1992) Non-
tion, and the equilibrium transport concentration, were dimensional parameters governing hyperconcentrated flow in an
given. The equilibrium transport concentration of the open channel. J Hydraul Eng-ASCE 36:221226 (in Japanese)
Iverson RM (1997) The physics of debris flows. Rev Geophys 35(3):
large particles that cannot be suspended by turbulence is
given uniquely as a function of the channel gradient. But, Iverson RM, Denlinger RP (1987) The physics of debris flows-a con-
for the smaller particles that are at least partly suspended ceptual assessment. In: Beschesta et al. (ed.) Erosion and sedi-
by turbulence, the equilibrium transport concentration mentation in the Pacific rim. IAHS publ. 165:155165
Chapter 5 Progress in Debris Flow Modeling 77

Iverson RM, Denlinger RP (2001) Flow of variably fluidized granu- Takahashi T (2004) Mechanics and countermeasures for debris flow.
lar masses across three-dimensional terrain. J Geophys Res Kinmiraisha, Nagoya (in Japanese)
106(B1):537552 Takahashi T, Nakagawa, H (1994) Natural dam formation and the
Johnson AM (1965) A model for debris flow. Ph.D. Dissertation Pa. disaster-a possible explanation of one extreme event. In: Proceed-
State Univ., State College ings International Workshop on Floods and Inundations related
Johnson AM (1970) Physical processes in geology. Freeman, New York to Large Earth Movements, A8.1A8.12
Julien PY, OBrien JS (1997) On the importance of mud and debris Takahashi T, Satofuka Y (2002) Generalized theory of stony and tur-
flow rheology in structural design. In: Proceedings 1st Inter-na- bulent muddy debris-flow and its practical model. J JSECE 55(3):
tional Conference on Debris-Flow Hazards Mitigation: Mechan- 3342 (in Japanese)
ics, Prediction, and Assessment, New York, ASCE, pp 350359 Takahashi T, Tsujimoto H (1997) Mechanics of granular flow in in-
Krieger JS (1972) Rheology of monodisperse lattices. Advance of Col- clined chute. J Hydraul Coast Environ Eng JSCE565:5771 (in
loid Interface Science 3:111136 Japanese)
Phillips RJ, Armstrong RC, Brown RA, Graham AL, Abbot JR (1992) Takahashi T, Nakagawa H, Satofuka Y (2000) Newtonian fluid model
A constitutive equation for concentrated suspensions that ac- for viscous debris-flow. In: Wieczorek GF, Naeser ND (ed) Debris
counts for shear-induced particle migration. Phys Fluids A4(1): flow hazards mitigation: mechanics, prediction, and assessment.
3040 Proceedings International Conference, Taipei, Balkema, Rotter-
Sabo Technical Center (1998) Reality of sediment disasters in 1997. dam, pp 255262
pp 2527 (in Japanese) Takahashi T, Nakagawa H, Satofuka Y, Kawaike K (2001) Flood and
Straub S (2001) Bagnold revisited: implications for the rapid mo- sediment disasters triggered by 1999 rainfall in Venezuela; a river
tion of high-concentration sediment flows. Spec Publs Int Ass restoration plan for an alluvial fan. J Natural Disaster Science,
Sediment 31:91109 JSNDS 23(2):6582
Takahashi T (1978) Mechanical characteristics of debris flow. J Hydr Takahashi T, Satofuka Y, Kashimoto S (2003) Motion of landslide-
Eng Div-ASCE 104:11531169 induced debris flow. In: Rickenmann D, Chen CL (eds) Debris
Takahashi T (1991) Debris flow. Balkema, Rotterdam flow hazards mitigation: mechanics, prediction, and assessment;
Takahashi T (1999) Mechanics of viscous debris flow. In: Takahashi Proceedings International Conference, Davos, Millpress, Rotter-
(ed) Japan-China joint research on the mechanism and the coun- dam, pp 399410
termeasures for the viscous debris flow. Research report of Group Tognacca C, Bezzola GR, Minor HE (2000) Threshold criterion for
C3 of special project associated with IDNDR sponsored by Min- debris-flow initiation due to channel-bed failure. In: Wieczorek
istry of Education, Science, Sports and Culture, Japan, pp 6484 GF, Naeser ND (ed) Debris flow hazards mitigation: mechanics,
Takahashi T (2000) Initiation and flow of various types of debris- prediction, and assessment. Proceedings International Confer-
flow. In: Wieczorek GF, Naeser ND (ed) Debris flow hazards miti- ence, Taipei, Balkema, Rotterdam, pp 8997
gation: mechanics, prediction, and assessment. Proceedings In- Yano K, Daido A (1965) Fundamental study on mud-flow. Bull Di-
ternational Conference, Taipei, Balkema, Rotterdam, pp 1525 saster Prev Res Inst Kyoto Univ 14:6983
Part II Landslide Dynamics

Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test

to Simulate Initiation and Post-failure Motion of Landslides

Chapter 7 Shear Behavior and Shear Zone Structure of Granular Materials

in Naturally Drained Ring Shear Tests

Chapter 8 Rockslides and Their Motion

Chapter 9 Residual Shear Strength of Tertiary Mudstone

and Influencing Factors

Chapter 10 On Failure of Municipal Waste Landfill

Chapter 11 Experimental Study with Ring Shear Apparatus

on the May 2004 LandslideDebris Flow
at Bettou-dani Valley, Haku-san Mountain, Japan

Chapter 12 On the Pore-pressure Generation and Movement

of Rainfall-induced Landslides in Laboratory Flume Tests

Chapter 13 Ring Shear Tests on Clays of Fracture Zone Landslides

and Clay Mineralogical Aspects

Chapter 14 Landslides Induced by a Combined Effect

of Earthquake and Rainfall

Chapter 15 Landslide Experiments on Artificial and Natural Slopes

Chapter 6

Undrained Stress-controlled Dynamic-loading Ring-shear Test

to Simulate Initiation and Post-failure Motion of Landslides

Kyoji Sassa* Hiroshi Fukuoka Gonghui Wang Fawu Wang

Abstract. Landslides are gravitational mass movements of rock, de- per shear box. This feature is particularly important for
bris or earth. Shear deformation in landslides before failure conforms the ring-shear apparatus as well as for the direct shear
to the field of statics. But shear deformation during seismic loading
apparatus, because it ensures an accurate determination
and post-failure motion of landslides conforms to the field of dy-
namics. Thus, study of the initiation of earthquake-induced land- of the total normal stress acting on the soil specimen dur-
slides and rapid landslide motion needs to develop Landslide dy- ing the experiment.
namics involving dynamic loading and dynamic generation/dissi- The apparatus developed by Bromhead (1979, 1986),
pation of excess pore-water pressure during motion. New develop- and by Savage and Sayed (1984), and the apparatus modi-
ments in science can be facilitated by new technological advances.
fied by Garga and Sendano (2002) do not have two (upper
This study aimed to develop a new testing method that can geo-
technically simulate the formation of the shear zone and the fol- half and lower half) sample boxes to accommodate shear.
lowing long and rapid shear displacement that occurs in high-ve- These apparatus have one sample box loaded and sheared
locity landslides. Sassa K and his colleagues at DPRI (Disaster Pre- by the upper loading platen. This mechanism is simpler than
vention Research Institute), Kyoto University, have worked to de- others because it is not always easy to prevent leakage of
velop an undrained stress-controlled dynamic-loading ring-shear samples through the gap between two rings. Shearing
apparatus and its testing method for this purpose. This paper de-
scribes the development of this testing method, and its application
takes place between the upper loading platen, bottom of
to the study of earthquake-induced landslides and landslide trig- the shear box and the soil sample. The mobilized shear
gered debris flows in Japan. resistance may differ from that mobilized during shear-
ing within the soil specimen, especially for sandy materi-
Keywords. Ring-shear test, landslide dynamics, earthquake-induced als; however, this probably poses no problem for clays.
landslides, landslide-triggered debris flows, undrained loading
The ring shear apparatus are also used for the study on
faults under much higher pressure and much smaller speci-
mens (Shimamoto and Tsutsumi 1994; Zhang et al. 1999).
6.1 Introduction The original high-speed ring shear apparatus (DPRI-1)
developed by Sassa (1984) used a conventional shear-
The ring-shear apparatus was designed initially to inves- speed control motor; it could not provide cyclic shear-
tigate the residual shear resistance mobilized along the stress loading. The first stress-controlled dynamic-load-
sliding surface at large shear displacements in landslides ing ring shear apparatus (DPRI-3) was developed to re-
because it allows unlimited deformation of the specimen. produce seismic loading using a torque-control motor and
The test configuration for the ring-shear device was in- a servo-control system that utilized the feed-back signal
troduced by Hvorslev (1939); Hvorslevs device forced the from a load cell that measured torque (Sassa 1994, 1995).
sample to begin shearing on a predefined plane located Following DPRI-3, a series of stress-controlled dynamic-
at the separation of the upper and lower confining rings. loading ring-shear apparatus has been developed that
This concept was utilized and improved by Bishop et al. incorporated different features (DPRI-4, DPRI-5, DPRI-6
(1971), Bromhead (1979), Savage and Sayed (1984), Sassa and DPRI-7 in Table 6.1).
(1984), Hungr and Morgenstern (1984), Tika (1989), and The general purpose of the DPRI ring shear testing
Garga and Sendano (2002). program is to design an apparatus that can quantitatively
The most well-known and widely adopted type of ring- simulate the entire process of failure of a soil sample, from
shear apparatus was developed jointly by scientists and en- initial static or dynamic loading, through shear failure,
gineers at the Imperial College of Science and Technology pore-pressure changes and possible liquefaction, to large-
(United Kingdom) and the Norwegian Geotechnical Insti- displacement, steady-state shear movement. No other
tute (Bishop et al. 1971). The main advantage of the Bishop laboratory apparatus has so far been able to provide an
type of ring-shear apparatus compared with other previ- integrated simulation of this natural process. The latest
ous models was provided by its ability to measure the fric- devices in the DPRI ring shear series have succeeded in
tion between the soil sample and the sidewalls of the up- this goal.
82 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

landslides. However, the current undrained ring shear

apparatus geotechnically simulates the formation of the
shear zone and the postfailure mobility of highspeed
landslides and observes the consequence of mobilized
shear resistance, as well as the post-failure shear displace-
ment and generated pore-water pressure.
Many unknown factors exist in landslide motion. To
simulate the natural sliding phenomena as clearly as pos-
sible, the following conditions were necessary, dictating
the new design features summarized in Table 6.1 in com-
parison with some previous ring shear apparatus:

1. Shearing is provided primarily by the stress-control

(more precisely by torque-control) condition because
the triggering factors in natural phenomena, such as
earthquake shaking, ground-water fluctuation, or
change of slope profiles due to toe erosion or filling,
Fig. 6.1. Design concept of the ring shear apparatus to simulate ini- are associated with changes in stress within the slope.
tiation and post-failure motion of landslides However, the speed-control test, which is the same as
that developed in the previous apparatus by Bishop
The mechanism of the undrained stress-controlled et al. (1971), can also be conducted to obtain the re-
dynamic-loading ring shear apparatus introduced in this sidual friction angle of the sample.
paper (Fig. 6.1) has basically the same principle as the 2. The maximum shear speed along the mid-diameter circle
ring shear device developed by Bishop et al. (1971) in of the specimen is 33 cm s1 (DPRI-3) to 300 cm s1
shearing soils for unlimited displacement; however, the (DPRI-7) to simulate rapid landslide motion.
major purpose and, consequently, the design concept is 3. The undrained condition under rapid shearing can be
rather different. The ring shear apparatus described by maintained by pressing rubber edges onto the bottom
Bishop et al. (1971) was designed to study the postpeak of an upper pair of shear boxes at a necessary contact
interval of the shear resistancedisplacement curve with pressure using a servo-oil piston and gap sensor with
emphasis on residual strength developed in slow clayey a precision of 1/1 000 mm.
Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test to Simulate Initiation and Post-failure Motion of Landslides 83

4. Pore pressure is monitored by a pressure gauge con- the shear stress needed to continue deformation and the
nected to the filtered gutter (4 4 mm) along the cir- velocity of deformation remain constant. As well known,
cumference of the inner wall of the upper outer ring although triaxial apparatus showed best achievement in
of the shear box. measuring the stress-strain relationship and maximum
5. Rapid loading and high-speed data acquisition are avail- shearing resistance, it is not suitable in determining the
able to simulate earthquake loading and other phenom- temporary or permanent decrease of the shearing resis-
ena involving rapid stress changes. A seismic wave form tance after failure (Bishop et al. 1971), due to the limita-
with frequency of up to 5-Hz can be reproduced, and a tion in shear displacement. Thus, the potential particle
data-acquisition rate from 12 readings per second (DPRI- orientation and particle breakage may be incomplete due
3) to 1000 readings per second (DPRI-7) is available. to the limitation of shear displacement, i.e., the steady state
6. The shear box is deep enough to allow for the develop- of deformation would be difficult, if not impossible, to be
ment of a well-defined shear zone within the soil speci- reached, say in triaxial tests, especially for those sands in
men in order to avoid the possibility of measuring the medium or dense state. In fact, in many cases, such as
shear resistance between soil specimen and the load- some landsides, the shearing behavior may involve runout
ing platen or the bottom of the shear box. of several meters to tens of kilometers, often on slopes of
7. It is also possible to conduct cyclic shear-displacement only a few degrees (Voight 1978, 1979; Rouse 1984). There-
controlled tests, torque-controlled tests, and shear- fore, quite large shear displacement is needed to examine
speed controlled tests in DPRI-4 and DPRI-7. these failure phenomena where the shearing resistance
may decrease along with shear displacement.
The undrained stress-controlled dynamic-loading ring To overcome this limitation in shear displacement and
shear apparatus were developed to accomplish the above measure the shear resistance after failure, ring shear ap-
conditions while reproducing the formation of the shear paratus had been developed and improved, and then
zone and the resulting motion along the shear zone. There- widely used in the analysis of slope stability (Bishop et al.
fore, the aim and design features are different from previ- 1971; Bromhead 1979; Gibo 1994; Tika and Hutchinson
ous ring shear apparatus. Table 6.1 presents the major 1999). However, almost all of the ring shear apparatuses
characteristics of some previous ring shear apparatus in used in these studies were not enable to perform undrained
comparison to the features of the undrained stress-con- shearing tests. Quite recently, a series of undrained ring
trolled dynamic loading ring shear apparatus developed by shear apparatuses was developed and improved by Sassa
Sassa and colleagues at the DPRI (Disaster Prevention Re- and his colleagues. These series newly developed undrained
search Institute), Kyoto University. ring shear apparatuses offered an available method to
study the undrained shear behavior of sands in virtually
limitless displacement levels.
6.2 Comparison with Triaxial Tests Difference from conventional ring shear apparatus is
not only undrained test capability, but capability of shear-
Most of geotechnical tests are conducted by triaxial tests. stress controlled tests. Shearing by conventional ring shear
Often question is raised on the difference. This question apparatuses is in the shear-speed controlled way (Bishop
especially focusing on the undrained shear behavior of et al. 1971; Bromhead 1979; Hungr et al. 1984; Garga et al.
sands is examined in this section. 2002). The series of stress-controlled dynamic-loading
Undrained shear behavior of sands is of great impor- ring-shear apparatus can conduct shear stress-controlled
tance to engineers in the anti-liquefaction designs. By now, test. The shear stress in any form, including monitored
a great number of experimental studies have been con- real seismic waves up to 5 Hz and an impact loading in a
ducted for a better understanding on the undrained shear way which a landslide mass falls on the torrent deposit,
behavior of sands. Based on the results of mostly triaxial can be loaded. Pore-pressure-control test to simulate the
tests, a concept of steady state was proposed and widely ground water rise in the slope is also conducted. There-
used in the procedure of analyzing liquefaction suscepti- fore, the apparatus is not to measure shear resistance as a
bility of a soil in a given geoenvironment (Castro 1969; parameter, but to reproduce the stress and pore pressure
Ishihara 1993; Vaid and Chern 1985; Poulos et al. 1985; acting on a potential sliding surface and the resulting slid-
Alarcon-Guzman et al. 1988; Kramer and Seed 1988). ing-surface formation by failure and the post-failure mo-
However, the determination of steady state is still a prob- tion. The generated excess pore pressure in the shear zone
lem needed to be scrutinized, because according to Poulos and the shear resistance mobilized on the sliding surface
(1981), steady state means the mass is continuously de- are monitored. Thus, the stress-controlled dynamic-load-
forming at constant volume, constant normal effective ing ring shear apparatus and its test is not a method to
stress and constant velocity. It is achieved only after all measure shearing parameter of an element of soil, but a
particle orientation has reached a steady state condition test to reproduce shearing phenomenon, sliding surface
and after all particle breakage, if any, is complete so that formation and post-failure motion, which will occur in
84 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

the natural slopes during rainfalls, earthquakes and other the figure; the servo-motor for shear-stress loading as
triggering stress changes. well as speed-control shearing is hidden behind the con-
Note that a comparison study on the undrained shear trol and monitoring unit.
behavior of granular material in ring shear tests and Figure 6.3 shows a schematic diagram of the appara-
triaxial compression tests has been performed and can tus, while Fig. 6.4 presents a brief illustration of the elec-
be obtained from Okada et al. (2000). trical control system of DPRI-5, 6, and 7. As shown in both
figures, the sample is loaded by a loading platen through
an oil piston (OP1), and the loaded normal force (for nor-
6.3 Structure and Control System of Apparatus mal stress) is measured by a load cell (N1). The sum of
the friction between the sample and the sidewall of the
6.3.1 Outlines upper pair of rings in addition to the self-weight of the
upper pair of rings is measured by a load cell (N2). The
Figure 6.2 presents the overall frontal view of DPRI-6 (the actual normal force acting on the shear surface is obtained
largest type within the DPRI ring shear family). The to- from the difference between N1 and N2. This value is sent
tal height is 5.1 m. The apparatus was installed in a pit at to a servo-amplifier as a feed-back signal (as shown in
the DPRI Landslide laboratory with the level of the Fig. 6.4). Then, the normal stress on the shear surface is
sample box approximately 1 m above the floor for easy automatically kept the same as the control signal given
access. The rear of the machine in Fig. 6.2 is the main by the computer. Shear stress is supplied by a torque-con-
apparatus, and the front instrument box is the comput- trolled servo-motor, which can be switched to a speed-
erized control and monitoring system. The shear box and controlled mode as well. The applied torque is measured
the oil piston for normal stress loading are indicated in by torque transducer T1. Using the monitored value of
T1 as the feed-back signal (Fig. 6.4), the applied shear
stress is automatically controlled by the servo-amplifier
and servo-motor and then kept the same as the pre-de-
termined value given by the computer. The shear resis-
tance acting on the shear surface is monitored by two load
cells (S1 and S2), through which the upper shear box is
restrained from rotation. The second shear-stress load-
ing system (servo-motor 2 and torque transducer T2) in
Fig. 6.3 is a spare because the manufacturing or repair-
ing of this special motor is a lengthy procedure.
Four computers are used in DPRI-6. One computer is
for controlling the test. The controlling signals could be
either edited signals of monotonic/cyclic loading, or seis-
mic records of earthquakes (as listed in Table 6.1). The
second computer monitors the response of each trans-
ducer, records the data, and then draws the effective-stress
path, simultaneously. Note that a multi-pen recorder is
also used for recording the data on paper. The third com-
puter is installed for monitoring the system functions and
providing the safeguard of an automatic alarm system. If
the apparatus is operated incorrectly or something hap-
pens to the mechanical parts, the alarm system will sound
the alarm and automatically terminate the test. The fourth
computer is mounted for memorizing the details of the
test procedures and data analysis.

6.3.2 Prevention of Water Leakage

The most essential part of the undrained ring shear ap-

paratus is the undrained shear box. Its design is illustrated
as Fig. 6.5, an enlarged diagram of the left half of the cross
Fig. 6.2. Photo of the undrained dynamic-loading ring shear appa- section of the undrained shear box and its surroundings,
ratus (DPRI-6) including the water-pressure measurement system.
Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test to Simulate Initiation and Post-failure Motion of Landslides 85

Fig. 6.3.
Mechanical structure of the ap-

The most difficult part is prevention of leakage of water shear box for DPRI-7 were made of transparent Acrylic.
through the gap between the upper shear box and the lower Acrylic is deformable under elevated temperature, a fac-
shear box during high-speed shearing (33300 cm s1). This tor taken advantage of in the assembly process. Periodi-
was perfected for DPRI-3 in 1992 and 1993 (Shoaei and cally, the tops of the rubber edges were leveled in the labo-
Sassa 1994). A material with strong abrasion resistance, ratory by pressing a hand file (fixed at a certain level) to
low friction characteristics and good performance as a both edges and rotating the lower rotary box. Before each
sealant had to be found. Shoaei and Sassa tried various test, the rubber edges were covered with a friction coat of
types and shapes of rubber edges, Teflon (polytetrafluor- Teflon (polytetrafluorethylene), and then daubed with
ethylene) plates, and combined layers of Teflon plate un- vacuum silicon grease.
derlain by rubber plate. The final selection was a stair- During the test, a certain amount of compression is
shaped rubber ring (rubber hardness index, 45JIS) as applied between the upper pair of rings and the rubber
shown in Fig. 6.5. High surface smoothness and equal edges by lowering the gap control oil piston (OP2 in
height of both the outer and inner rings were found to be Figs. 6.3 and 6.4) to a specific value (hereinafter termed
extremely important. After pasting rubber edges on the as Gap Value). To maintain the undrained condition, con-
two confining rings of the lower rotary box, rubber was tact pressure between the rubber edges and the upper pair
cut by a lathe and then polished to a smooth surface in of rings should be kept greater than the generated pore
the assembly process. The outer confining rings of the pressure in the sample. Then during the test, an effort is
86 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

made to keep this compression force constant by main-

taining the Gap Value as constant as possible by means of
an oil piston (OP2), using the feed-back signal obtained
from a gap sensor (GS) with a precision of 1/1000 mm.

6.3.3 Pore-pressure Monitoring

Development of an effective and durable pore-pressure

monitoring system was also a difficult task. At first a
needle that was connected to a pore-pressure transducer
was inserted to near the shear zone from the upper load-
ing platen (Sassa 1995). However, the needle was deformed
during shearing, and did not have large enough inlet sec-
tion for the pore-pressure transducer. The transducer has a
diaphragm that is deformed by water pressure to provide
electrical output. To have a large inlet section and provide
an average pore-pressure value throughout the soil sample,
pore-pressure transducers are connected to a gutter
(4 4 mm) extending along the entire circumference of the
inner wall of the outer ring in the upper box, as shown in
Fig. 6.5. The gutter is located 2 mm above the shear surface
and is covered by two metal filters, with a filter cloth be-
tween them. This system is quite durable in regard to shear-
ing and is sensitive to pore-pressure monitoring, although
Fig. 6.4. Electronic Control System the monitoring point is not at the center of the shear zone.

Fig. 6.5.
A half section of the shear box
and the close-up diagram of the
Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test to Simulate Initiation and Post-failure Motion of Landslides 87

6.3.4 Efficiency of Servo-control System 6.4 Testing Procedure

and Water Pressure Measurement System
The test procedures could be separated as two parts, namely
The efficiency of servo-control system and water pres- preparation test and test on saturated sand. The purposes
sure measuring system was evaluated by conducting cy- of the preparation test are adjusting the Gap Value to give
clic loading (normal stress) on water within the undrained an appropriate pressing force between the upper pair of
shear box. During test, the target cyclic loading was de- rings and the rubber edges, at which time there will be no
signed and input in the controlling computer, while the water leakage during undrained shearing tests; and then
output response was checked by monitoring the response measuring the friction between the rubber edge and the
of pore water pressure transducers. The time series data of upper parts of ring box corresponding to the adjusted gap
loaded normal stress are shown in Fig. 6.6a, while the mea- value. The tests procedures are listed below.
sured response of pore water pressure transducer is plotted
in Fig. 6.6b. Figure 6.6c presents these two components
against time in one cycle. It is seen that essentially the 6.4.1 Preparation Test for Adjustment and Correction
loaded normal stress behaved the same as the monitored
pore water pressure, showing the compatibility of con- Gap Value Adjusting
trolling system and water pressure measuring system.
As mentioned above, there are many kinds of shearing An appropriate Gap Value is necessary; too small press-
methods using DPRI-Ver.6. The following section will ing force could not ensure water leakage proof, while too
introduce the basic operating methods by performing great pressing force would cause damage to the rubber
undrained shearing tests under monotonic loading con- edges and give a greater rubber edge friction. After the
dition. To observe the shear behavior of soil accompany- rubber edge was cleaned, covered with Teflon, daubed by
ing the increasing shear stress before failure, torque-con- silicon grease, the apparatus was set up without sample.
trolled method is selected. And then, the upper parts (including the upper pair of
rings and the cap plate shown in Fig. 6.5) were raised by
increasing the Gap Value until the upper parts were sepa-
rated completely from the rubber edges and the reading
of the amplifier for N2 in Fig. 6.2 did not change anymore,
while the loading platen was kept free. Thereafter, the
amplifier for N2 was adjusted to zero, and the upper parts
were pulling down by lowering the oil piston of OP2 until
a certain Gap Value, at which the reading of N2 (which is
the Gap Contact Force caused by the touching and press-
ing of the upper pair of rings on the rubber edges) reached
a certain value (determined according to different test
conditions, and was kept as approximately 250 kgf in the
tests presented in this report). If this contact pressure
between the rubber edges and the bottoms of upper pair
of rings is always kept greater than the generated pore pres-
sure inside the sample along the whole rubber edge contact
surface, any leakage of water should not occur. The rubber
edge contact surface is precisely processed to be smooth
by machine after being pasted onto the lower parts. The
Gap Value was kept constant by the servo-oil piston in
the commenced undrained shear test on saturated sample,
while the readings in the amplifier for N2 (250 kgf) was
adjusted to zero before the loading of normal stress so to
give a pure reading for the total value of friction between
the sample and the sidewalls of upper rings.

Fig. 6.6. Response of the loading system and monitoring system in Leakage Proof Checking
the case of loading on only water in the shear box. (a) and (b) show
10 cycles of the applied normal stress and the resulting monitored
water pressure; (c) shows one cycle of applied normal stress and the Water leakage proof was checked by putting water into
resulting water pressure (pore pressure) the shear box and applying normal stress on water and
88 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

value (mobilized maximum static friction at Point Fp) due

to the fact that friction was shifted to kinetic friction af-
ter Point Fp, and thereafter increased with progress of
shear displacement, probably due to the abrasion of the
covered Teflon and the daubed silicon grease with
progress of shear displacement; finally, it became almost
a constant. This measured rubber-friction with shear dis-
placement was subtracted from the monitored shear
strength for the correction of shear strength.

6.4.2 Test on Saturated Sand Sample Setting

The samples were made by means of moist placement or

dry deposition (Ishihara 1993), according to different test
purposes. For moist placement method, de-aired water
Fig. 6.7 Checking the water leakage proof at the adjusted gap value was first added to the oven-dried samples to make the
initial water content rise up to 5 percent, and then the sand
was stirred evenly. After that, the sample was placed into
the shear box. To make the sample uniform, while pack-
ing, the sample was placed in a series of layers of 3 cm
thickness, and then each layer was tamped. For dry depo-
sition method, the oven-dried sample was fallen into the
shear box freely by layers, and each layer was and was not
tamped differently to make the initial density different. Sample Saturation

Sample was saturated with help of carbon dioxide and

Fig. 6.8. Friction of rubber edge against shear displacement (shear de-aired water. After the sample was packed, CO2 was then
speed = 10 mm s1) percolated through the sample to expel the air in the
sample pores out, by flowing in through the lower drain-
shearing, while keeping the Gap Value as the same as ad- age line very slowly, and discharging from the upper drain-
justed. If there is no decrease in the water volume within age line. Usually, this process took 4 to 12 hours, depend-
the shear box during rotation under the loaded normal ing on the samples (for the sample used in this research,
stress, it is ensured that there is no water leakage using 4 hours are enough). After hours of percolation of CO2,
the adjusted Gap Value. Figure 6.7 presents the applied de-aired water was infiltrated into the sample through the
normal stress, monitored water pressure, and the volume lower drainage line to expel the CO2 in the sample pores
change (in the form of sample height) against the time. As from the upper drainage line. This infiltration was kept
shown, no change occurred in them with time, showing the very slow with help of a very small water head. To expel
efficiency of leakage proof under the adjusted Gap value. the CO2 as completely as possible, this water saturation
process was usually kept 24 hours or longer. Rubber Friction Measuring Saturation Checking
After Leakage proof checking was finished, the friction
between the upper pair of rings and the rubber edges was Saturation degree was checked by using BD parameter. BD
measured. In this apparatus, there are three kinds of ro- is a parameter of saturation degree in the direct shear state,
tating gear with final speed of Low (10 mm s1), Medium which was proposed by Sassa (1988), and formulated as:
(32.3 cm s1) and High (2.24 m s1). In this test, the Low
gear was selected. Figure 6.8 presents the measured rub- BD = u/ (6.1)
ber friction corresponding to the adjusted Gap Value at
the shear speed of 10 mm s1 of the Low Gear. As seen, where u and are the increment of pore pressure and
the friction showed a transient reduction after the peak normal stress, respectively. During checking, the sample
Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test to Simulate Initiation and Post-failure Motion of Landslides 89

was firstly consolidated under normal stress of 49 kPa relating to the undrained shear behavior (Hungr and
(0.5 kgf cm2) in drained condition. Thereafter, a normal Morgenstern 1984). After consolidation, undrained shear
stress increment, = 49 kPa, was applied in undrained stress was subsequently applied at a loading rate of
condition, and the resultant increment of excess pore pres- 0.098 kPa s 1 (0.001 kgf cm 2 s 1). Transducers were
sure (u) was measured. And then, the saturation degree scanned at an interval of 1 second before the peak shear
was checked by the ratio (BD) of excess pore pressure in- strength; after that, the sampling rate was increased to
crement and normal stress increment (u/). Because 20 samples per second because of rapid shearing, then
high saturation degree is necessary for acquiring correct decreased again to 1 sample per second when the change
monitoring data, in this study, all the tests were carried in each parameter became not so quick (Note that a
out with BD 0.95. greater sampling rate on 200 samplesper second is used
for seismic loading or other faster phenomena). Sample Consolidation Sample Characteristics
All the samples were normally consolidated in this series
tests under the pre-decided normal stress and shear stress. In this test, silica sand no. 8 (S8) was selected as the sample.
After the checking of BD, the normal stress was decreased Silica sand is a kind of sand material for building use made
in undrained condition to a value (usually smaller than from silica sandstone by grinding, comprised of suban-
49 kPa, due to the plasticity deformation nature of saturated gular to angular quartz. The particles ranged from fine
sand) where the pore pressure decreased to zero, and then sand to silt sizes with a specific gravity of approximate
the upper drainage valve was switched to open. Thereafter, 2.63. The maximum and minimum dry densities are found
normal and then shear stresses were loaded slowly to the to be approximately 1.42 and 0.99 g cm3, respectively. The
decided values. In the present research, all the tests were grain size distribution is presented in Fig. 6.9. As shown,
carried out under the same initial stress state, with normal the mean diameter is approximately 0.05 mm.
stresses being 196 kPa and shear stress being 0. Undrained Shearing 6.5 Undrained Shear Behavior on Sands

The studies by both Castro et al. (1982) and Sladen et al. Since the trust of this note is on the introduction of the
(1985) have made it clear that there is no difference be- design of the undrained ring shear apparatus and its op-
tween the results of either load (stress)-controlled or erating procedures, here only selected typical results from
strain-controlled tests. The torque-controlled method was the tests on loose and dense sands are presented. More
selected to take more data between the start point of shear- detailed discussion of the undrained behavior of sands
ing and the point where peak shear strength was mobi- in ring shear tests will be presented in a separate paper.
lized, and then to obtain a well-defined effective stress
path until a steady state after a long shear displacement
of tens of meters. To avoid the phenomenon of quick 6.5.1 Undrained Shear Test on Loose Sand
motion in contractive sand once the peak shear stress has
been reached, which usually appears in stress-controlled The results of one test on loose silica sand S8, showing
triaxial compression tests, in this test, the Low gear (as the same effective stress path as that of liquefaction be-
mentioned above) was used for careful monitoring, pro- havior reported for many undrained triaxial tests on loose
viding that shear speed has no effects on the key results sands, are presented in Fig. 6.10d. The specimen for this
test was set by means of moist placement and consoli-
dated under the normal stress of 196 kPa. After consolida-
tion, the relative density reached 63.3% (1.15 in void ratio).
Figure 6.10a shows the variation of normal stress and
pore pressure in relation to shear displacement, and
Fig. 6.10b superimposes the monitored shear resistance
and rubber friction (shown in Fig. 6.8) as well as the cor-
rected shear resistance of sand against shear displacement.
Figure 6.10c plots the time series data of normal stress, cor-
rected shear resistance and pore pressure. In Figs. 6.10a and
6.10b, to make a clear view on the generation of pore pres-
sure accompanying the shear displacement in the initial
shearing period, a logarithmic abscissa of shear displace-
Fig. 6.9. Grain size distribution of silica sand no. 8 (S8) ment within the range of 0.1 m was taken, and thereafter,
90 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

linear abscissa was used to show that the test had been ing was completed (reached at point SSP in Figs. 6.10a and
sheared to a steady state (point SSP in Fig. 6.10a). It is 6.10d), the upper drainage line was switched to drained
seen that some pore pressure was built-up before the peak condition so that the generated pore pressure could dis-
shear strength (point F in Fig. 6.10b), while after failure, sipate, while the low part of ring shear apparatus was kept
pore pressure showed a sharp increase, and shear resistance rotating at a constant speed. The stress shifted from SSP
underwent a quick reduction in terms of time (dropped to RS1, when the generated pore pressure dissipated com-
from F to P in Fig. 6.10c). This period is usually known as pletely. Then, the loaded normal stress was reduced at a
the collapse period, mainly due to the failure of meta- very small rate under drained condition while the shear
stable structure. After point P, accompanying the further resistance was measured. The stress moved from RS1 to
shearing, pore pressure built up gradually and then shear RS2. The line connecting RS1 and RS2 shows the residual
resistance decreased slowly as the subsequence. failure line (R.F.L.) of this sample. As shown in Fig. 6.10d,
Figure 6.10d presents the effective stress path and fail- soon after the start of shearing, with increasing shear
ure line. The residual failure line (R.F.L.) was measured af- stress, stress path moved towards but did not reach the
ter the undrained shearing test; after the undrained shear- failure line until the final point, i.e., the steady state. This
effective stress path showed the same changing tendency
as those in undrained triaxial tests on loose sand.

6.5.2 Undrained Shear Test on Dense Sand

Figure 6.11 shows the results of a test on dense sand. This

sample was made by means of dry deposition with heavy
tamping. After saturation and normal consolidation, the
sample was sheared in undrained condition to a large dis-
placement of approximate 73 m. Figure 6.11a plots the
normal stress and pore pressure against shear displace-
ment, while Fig. 6.11b shows the variation of the moni-
tored shear strength, rubber friction, and the corrected
shear resistance of sand in relation to shear displacement.
Note that the rubber friction shown in Fig. 6.8 was 7.3 kPa
at the terminal shear displacement of 46 m, and was
treated as the same as 7.3 kPa until the final shear dis-
placement of 73 m in this test. Figure 6.11c presents the
time series data of these components. As shown in
Fig. 6.11a, in the initial period after undrained shear stress
was applied, with increasing shear displacement, pore
pressure built up gradually. However, after point PT
(Phase Transformation), pore pressure decreased due to
the dilatancy of dense sand. After the peak shear strength
was reached (Point F in Figs. 6.11b and 6.11c), sample
failed, and thereafter, pore pressure built up with shear
displacement, finally reached 135 kPa approximately. The
shear resistance decreased consequently, and finally fell
to a constant of approximately 53.5 kPa (as seen clearly
in Fig. 6.11c). The excess pore pressure ratio, which is
determined as the ratio of excess pore pressure and ini-
tial effective normal stress, was approximately 0.69.
The effective stress path is shown in Fig. 6.11d. Upon
increase of shear stress, the effective stress path extended
Fig. 6.10. Ring shear test on loose sand (BD = 0.99, Dr = 63.3%, left-upward due to the pore pressure generation. After
= 196 kPa). a Variation of normal stress and pore pressure with point PT, the path went right-upward accompanying
shear displacement; b monitored shear resistance, rubber edge fric-
tion and corrected shear resistance against shear displacement; c time
further shearing, showed a shape of elbow with a turn-
series data of normal stress, pore pressure and corrected shear re- ing point. After failure point F, the path fell downward
sistance; d effective stress path until a small shear stress along the residual failure line.
Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test to Simulate Initiation and Post-failure Motion of Landslides 91

The stress path before the point F in Fig. 6.11d was

6.6 Geotechnical Simulation of
described as same as typical dilative behavior of the
Earthquake-induced Landslides
undrained shear behavior in triaxial apparatus (Castro
1969). Due to the limitation of triaxial apparatus in shear
displacement, the behavior after F was not usually ob- Figure 6.12a shows the stress conditions in a slope, where
tained. This phenomenon of large stress reduction along the weight of the soil column (W) and the dynamic stress
the failure line (from F to SSP) could not be reproduced well (which can be expressed by kW) are acting along a po-
before the development of undrained ring shear apparatus tential sliding surface. The stress path due to dynamic
by Sassa and his co-workers, and it is termed as sliding stress loading is expressed by TSP in terms of total-stress
surface liquefaction, due to the grain crushing with progress and ESP in terms of effective-stress in Fig. 6.12b. TSP is
of shear displacement (Sassa 1996; Sassa et al. 1996). determined automatically, but ESP depends on excess
pore-pressure generation during stress loading, as well
as during motion. Generally speaking, dynamic stress on
slopes is not limited to seismic stress and the direction of
seismic stress is not constant in the natural condition. So
the loading angle () and the seismic coefficient (k) are
time functions in nature.
Earthquake-induced landslides often move rapidly,
have long paths of travel, and cause catastrophic disas-
ters. For these reasons, a project called APERIF (Areal
Prediction of Earthquake and Rainfall Induced Rapid and
Long-traveling Flow Phenomena) was launched by the
Special Coordinating Fund for Science and Technology
of the Ministry of Education, Culture, Sports, Science and
Technology (MEXT) of Japan. In 2002, this project was
approved as one of International Programme on Land-
slides (IPL M101-APERITIF Project of the International
Consortium on Landslides).
As a part of this project, the upper slope of the Nikawa
landslide site, at which 34 people were killed as a result of
the 1995 Hyogo-Ken Nanbu earthquake (Sassa 1996; Sassa
et al. 1996), was investigated to determine whether or not
this slope could undergo retrogressive rapid landslide
activity if a possible similar earthquake were to occur.
Figure 6.13 is a geological cross section through the slope
as estimated by geological drilling. Within the slope, we
found a geological contact that can be a potential sliding
surface. The angle and shape of the slope are very similar
to those of the 1995 Nikawa landslide. There is a granitic
sand layer above and a gravel layer below the potential
sliding surface. The gravel layer includes silt with low
permeability. Samples were taken from both layers and
laboratory seismic simulation tests were performed us-
ing the ring shear apparatus. The details can be found in
Sassa (2002) and Sassa et al. (2003). During the tests, the
applied seismic loadings acting on the potential sliding
surface during the earthquake were synthesized by using
the seismic waves that were recorded at the Japan Rail-
way Takarazuka station (7 km from the earthquake fault)
Fig. 6.11. Ring shear test on dense sand (BD = 0.99, Dr = 95.2%, during the Hyogo-Ken Nanbu earthquake. This synthe-
= 196 kPa.). a Variation of normal stress and pore pressure with sis took into account attenuation laws for horizontal peak
shear displacement; b monitored shear resistance, rubber edge fric-
tion and corrected shear resistance against shear displacement; c time
acceleration proposed by Fukushima and Tanaka (1990)
series data of normal stress, pore pressure and corrected shear re- and vertical peak acceleration proposed by Ambraseys
sistance; d effective stress path and Bommer (1991), as well as the amplification of seis-
92 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

Fig. 6.12.
Stress conditions in a slope and
in the ring shear apparatus

Fig. 6.13. Cross section through the Nikawa slope

mic waves (usually 1.4 times) when these waves transfer

from hard rock to a weak soil layer (Fukushima and
Tanaka 1990). The synthesized seismic loadings are pre-
sented in Fig. 6.14.
The seismic loadings shown in Fig. 6.14 were applied
to samples taken from the granitic sand layer and from
the silty gravel layer. These samples were placed and nor-
mally consolidated in the ring shear apparatus. The re-
sults of tests on the sample of the granitic sand layer un-
der undrained conditions are presented in Fig. 6.15. The
monitored test results show that the normal stress ex-
pected in Fig. 6.14 was approximately reproduced inside
the apparatus, while the mobilized shear resistance was
less than the given shear stress because the shear stress
reached the failure line and failed there. In other words, Fig. 6.14. Synthesized seismic loadings on the inferred potential slid-
failure occurred. Shear displacement began during the ing surface
main shock and increased its speed during the post-fail-
ure shear process. The total stress path (TSP) is shown in apparent friction angle calculated from the arc-tangent
blue, its initial stress corresponding to the stress acting of the mobilized shear resistance divided by the total nor-
on the potential sliding surface at a depth of 26 m with an mal stress was about 3.5 degrees. Therefore, it was esti-
inclination of 20 degrees. The effective-stress path (ESP) mated that a rapid landslide could be initiated retrogres-
is shown in red. The initial difference between TSP and sively if an earthquake with the same magnitude and seis-
ESP corresponds to the ground-water level that is 16 m mic waveform as the 1995 Hyogo-Ken Nanbu earthquake
above the potential sliding surface. During seismic load- were to occur in the future under the undrained condition.
ing, the stress path deviated considerably because pore- A key assumption in this test is that a sliding surface
pressure monitoring did not always follow the rapid load- can be formed passing through a normally consolidated
ing. However, it was found that the ESP reached the fail- part, or a well-softened part similar to a normally con-
ure line, and then reduced along the failure line until it solidated state. Observation of cores of five drillings in
reached a certain low steady-state stress. The mobilized this site showed that the sand and silty gravel layers were
Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test to Simulate Initiation and Post-failure Motion of Landslides 93

Fig. 6.15. Undrained cyclic tests on saturated sand layer (BD = 0.99,
initial dry density = 15.2 kN m3). a Time series data for normal stress Fig. 6.16 Aerial oblique view of the 2003 Minamata debris flow
and shear stress; b time series data for pore-water pressure and shear
displacement; c stress path
6.7 Slide-triggered Debris Flow
disturbed by tectonic, seismic, and possibly gravity
stresses in the past. Therefore, some parts of the slope On 20 July 2003, a landslide occurred in an andesitic
are disturbed, namely well-softened, and some parts are weathered lava layer on a mountain slope of 3132 de-
over-consolidated and stiff. While conducting slope sta- grees in Minamata City, Kumamoto Prefecture, Kyushu
bility analysis in the course of planning for works to pre- Island, Japan. It was triggered by a heavy rainstorm with
vent a possible sliding surface liquefaction phenomena 314 mm total rainfall and a maximum rate of rainfall of
in this site, it would be dangerous to assume that the slid- 91 mm hr1. The slide mass entered a torrent, where it was
ing surface will pass into the stiff part of soil layer. A more transformed into a debris flow that struck a village along
prudent assumption is that the potential sliding surface the torrent, destroying 15 houses, killing 15 people, and
will be formed in the normally consolidated layer. This injuring an additional six people. Figure 6.16 presents a
assumption was adopted in the present study. view of the debris flow. The initial slide can be seen at the
In the APERITIF project, a series of naturally drained head of the debris flow. Apparently, this debris flow was
tests was also conducted, where the upper valve of the shear triggered by the slide, and the landslide mass flowed
box was kept open during the test. Because the permeabil- downstream along the torrent, increasing its volume by
ity of the sands and gravels was not small, generated pore- entraining material from the channel and weathered sur-
pressure could be dissipated partly during the short time face soils of the mountain slopes on both sides of the chan-
period of the main shock. This type of tests could be more nel. Figure 6.17 shows the central section through the ini-
practical. A mitigation measure for the prevention of land- tial slide surveyed by a non-mirror total station, and
slides by lowering the groundwater to cancel out the pore- Fig. 6.18 is a photo presenting the sampling point of the
pressure generation during earthquake loading has been weathered andesitic lava in the source area. Based on a
proposed (Sassa et al. 2003). Quantitative estimation of topographic survey made after the landslide occurred, the
pore-pressure generation during earthquakes is necessary initial slide was estimated to have occurred along a fail-
for reliable prediction of earthquake-induced landslides. In ure surface with an inclination of 26.5 degrees and depth
this regard, this series of ring shear tests provided an effec- of approximately 1012 m.
tive approach for the prediction of earthquake-induced Two samples were collected, one from andesitic lava,
landslides and design of the remedial measures. one from tuff breccia. The andesitic lava sample was taken
94 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

under the sliding surface near the head scarp at the mark the upper valve. The test result is shown in Fig. 6.19. Af-
of Sample A in Fig. 6.17. The tuff breccia sample was col- ter the stress path reached the failure line, it suddenly
lected from the mountain slope near the channel in the dropped to a much lower value. Thereafter, the shear re-
flank of the initial landslide at the mark of Sample B (pro- sistance slightly recovered to a certain value.
jected to the central cross-section) in Fig. 6.17. This ini- This action of rapid drop of shear resistance was in-
tial landslide occurrence was geotechnically simulated terpreted as follows: when shear failure occurred, volume
using the DPRI-5 ring shear apparatus. The initial stresses reduction took place due to grain crushing and the re-
on the sliding surface were reproduced in the apparatus;
then, the pore pressure was gradually increased simulat-
ing the rise of ground-water level during rainfall. In natu-
ral slopes, the ground-water rise would not be rapid; thus,
the undrained loading condition was not used. To simu-
late drained ground-water conditions, water pressure sup-
plied to the shear box through the upper drainage valve
was gradually increased. Thus, water pressure was con-
trolled, but the water was free to move through the upper
valve. Therefore, the sample was subjected to a natural
drained condition. If pore pressure was generated in the
shear zone during loading, it drained naturally through

Fig. 6.17. Central section through the Minamata landslide in its origi- Fig. 6.19. Results of tests to simulate the initiation of the Minamata
nal state landslide (BD = 0.86)

Fig. 6.18.
Sampling of weathered andesitic
lava in the Minamata landslide
Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test to Simulate Initiation and Post-failure Motion of Landslides 95

sulting failure of the soil structure; this caused rapid ex-

cess pore-pressure generation. However, excess pore pres-
sure was not monitored by the pore-pressure transducer
because the upper valve was open and the transducer was
not located at the shear zone, as illustrated in Fig. 6.5.
Therefore, the stress path deviated from the failure line
and dropped vertically.
The high pore pressure in the shear zone dissipated
through the upper valve, but some excess pore pressure
remained as the combined effect of pore-pressure gen-
eration within the shear zone and pore-pressure dissipa-
tion from the shear zone. The pore pressure generation
speed was likely to have decreased from the immediate
post-failure condition, while at the same time, the pore pres-
sure dissipation speed was maintained. Therefore, the shear
resistance recovered somewhat and settled to a certain value.
The result gave an apparent friction angle of 9.7 degrees.
This test is called a naturally drained test because gen-
erated excess pore pressure is naturally drained from the
shear zone to the upper valve. After the naturally drained
test, the sample was again consolidated by dissipating the
generated pore pressure. Then, slow shearing was applied.
After reaching the failure line, the normal stress was de-
creased very slowly to maintain the drained condition,
continuing slow movement without generating any pore-
water pressure. This test determined the status of the fail-
ure line during motion, as shown by the blue line in
Fig. 6.19. This test gave 32.9 degrees as the friction angle
during motion.
These test results show that the andesitic lava deposit
Fig. 6.20. Model of the landslide triggered debris flow (Sassa et al.
was subjected to high excess pore-pressure generation due 1997). a Illustration of the model; b stress path of the torrent de-
to post-failure shearing and a low apparent friction angle posit during loading. : angle of thrust between the slope and the
of 9.7 degrees was mobilized. This means that the Minamata torrent bed; Fd: dynamic stress; kd: dynamic coefficient (FD/W)
slide rapidly reached the torrent bed. The gradient of the
torrent bed at the toe of the slope was 17 degrees; so, the The stress working on the bottom of the soil column is
landslide necessarily continued to move downstream and presented in Fig. 6.20b. The initial stress is expressed by
changed into a debris flow. the point A, which corresponds to the position (I) of the
Loading by the failed slide mass onto the pre-existing sliding mass in Fig. 6.20a. If no excess pore pressure is
torrent deposits was first modeled by Sassa et al. (1997) generated during loading, the stress point moves to
as shown in Fig. 6.20. The slide mass moved down the point (C) by adding the static stress (W) to the initial
slope (I), and applied load onto the torrent deposits at stress. In addition, by adding the dynamic stress (Fd) to
the foot of the slope (II). Because a surface water stream the static stress, the total stress moves to point (B). There-
or subsurface flow existed and some of the deposits were fore, the stress path in the actual field case tends to move
saturated, the torrent deposit was sheared by undrained from point A to point B. However, when the stress path
loading and transported downstream together with the reaches the failure line, it moves along the failure line as
sliding mass (III). Let us consider a column of unit width, seen in Fig. 6.20b, because the stress path cannot exceed
which is a part of the torrent deposit. In the position (I) the failure line. At the point at which the dynamic stress
of the sliding mass, the weight of the column (W0) was in reduces to zero, the total stress moves back to the stress
effect. When the sliding mass rode on to the torrent point (C), namely the sum of W0 and W. Denoting the
deposit (II) with a certain velocity, it provided dynamic angle of thrust at collision with the torrent deposit as
loading of the column. Here, we assume that the applied and the dynamic stress as Fd, using a dynamic coefficient
stress on the torrent deposits was as the sum of the static kd = (Fd/W), the dynamic shear stress and normal stress
stress, W, (load due to the weight of the sliding mass) are expressed as:
and the dynamic (impact) stress, Fd, working in the di-
rection of motion of the sliding mass. Fd cos = d , Fd sin = d (6.2)
96 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

The stress path from A to B to C is the total-stress path of andesitic lava or tuff breccia. The surface soil layers of
in the case where no pore-pressure is generated. How- mountain slopes near the upper and middle part of tor-
ever, excess pore pressure is likely to be generated during rent was mainly tuff breccia, though the lower part of the
loading and also during shearing after failure. In this case, torrent was composed of non-volcanic sedimentary rocks.
the effective-stress path will deviate from the total-stress Therefore, two undrained loading tests were conducted
path as a curved line from A to D. for deposits of tuff breccia and those of andesitic lava.
When the landslide mass moves from the steep slope The test result of Fig. 6.22 shows the case for tuff brec-
to a gentle slope, the angle in Fig. 6.20 is great, but when cia, in which the torrent deposit and/or the surface soils
the landslide mass (i.e., the debris flow) travels along the were sheared and moved together with the original slide
torrent (as shown in Fig. 6.21), the angle is zero. An- mass. Only 18.5 kPa was necessary as additional shear
other test was conducted to simulate the landslide (de- stress to cause shear failure and the mobilized apparent
bris) masses as it moved onto the torrent deposits and/or friction angle was only 1.9 degrees in this rapid and
the surface soil layer of both slopes assuming the depth undrained loading condition as estimated from Fig. 6.22b.
of moving mass was about 10 m, the gradient of the tor- The test result of Fig. 6.23 shows the case for andesitic
rent bed was 15 degrees, the depth of the torrent deposit lava. This is stronger, the additional necessary shear stress
or surface soils was 24 m, and the dynamic coefficient was 45.5 kPa and the mobilized apparent friction angle
was 0.9. Because of such rapid loading by the fast-mov- at the steady state was 3.4 degrees. Both cases suggested
ing slide mass (more than 10 m s1), the test was carried that such saturated deposits were scraped and included
out under undrained conditions. The sliding surface was in the moving mass. However, if the material was not fully
formed inside the torrent deposits which were composed saturated, it probably would not be scraped as the case

Fig. 6.21.
Illustration of moving landslide
mass along a torrent

Fig. 6.22. Results of a test to simulate undrained loading of the tuff Fig. 6.23. Results of a test to simulate undrained loading of the weath-
breccia deposits in the Minamata landslide (BD = 0.89) ered andesitic lava in the Minamata landslide (BD = 0.97)
Chapter 6 Undrained Stress-controlled Dynamic-loading Ring-shear Test to Simulate Initiation and Post-failure Motion of Landslides 97

study in the Kameyama landslide, Hiroshima, Japan In addition, the volume of the landslide mass was in-
(Wang, Sassa and Fukuoka 2003). creased by entrainment of these deposits/soils. A new
To examine and interpret the actual phenomena, ad- ring shear apparatus (DPRI-7) with a transparent shear
ditional investigation and testing are necessary. However, box was developed in 2003. It enabled to observe the
this test has shown that this apparatus and its application shear zone during the initiation and post-failure motion
provide an effective tool to study the mechanism of slide- of landslides. DPRI-7 and the test results are introduced
triggered debris flows and the increase of landside mass in other papers (Sassa et al. 2004; Fukuoka et al. 2006;
by material entrainment during the flow. and in this volume).

6.8 Summary and Conclusions Acknowledgments

This paper has presented the principles of design and The first attempt at design and fabrication of an undrained
construction of a series of undrained ring shear appara- stress-controlled dynamic loading ring shear apparatus was
tus and its testing method developed and improved by conducted with the support of Scientific-Grant-in-Aid
Professor Sassa and his colleagues at DPRI, Kyoto Uni- (No. 03556021) of the Ministry of Education, Science, Cul-
versity. These test machines enable three kinds of shear ture and Sport of Japan in 1992 (Sassa 1994). This appara-
control, namely, the torque-, speed- or shear-displace- tus was designated as DPRI-3. The system and the testing
ment-controlled methods. But natural phenomena are procedures were improved during Ph.D. studies by Dr.
under the stress control condition. Therefore, stress con- Zieaoddin Shoaei, and by others at DPRI, Kyoto University.
trol tests to reproduce natural stress state during earth- DPRI-5 and DPRI-6 were developed by means of financial
quakes and rainfalls were introduced. The controlling sig- support from the Ministry of Education, Science, Culture
nals can be either edited signals of monotonic/cyclic load- and Sports of Japan, for earthquake-disaster mitigation re-
ing or seismic records of earthquakes. search after the Hyogo-Ken Nanbu earthquake of Japan in
Introduction of a rubber edge and the gap controlling 1995, Japan. Apparatus DPRI-7 was developed with sup-
system enabled the system to be leakproof even when the port from the project, Areal Prediction of Earthquake and
sample was sheared at high speed (i.e., at the maximum Rain Induced Rapid and Long-traveling Flow Phenomena
rotating speed of 224 cm s1 for DPRI-6 and 300 cm s1 (APERIF), of the Special Coordinating Fund for Promoting
for DPRI-7). This leakproof system enabled examination Science and Technology of the Ministry of Education, Cul-
of the undrained shear behavior of soils with high mobil- ture, Sports, Science and Technology of Japan (MEXT).
ity at essentially limitless shear-displacement levels. APERIF project was approved as a project of the Inter-
The assessment of landslide risk of the Nikawa slope national Programme on Landslides (IPL) by the Interna-
area was performed on the basis of field investigation and tional Consortium on Landslides (ICL) as IPL M-101
laboratory ring shear simulation tests. Field investigation APERITIF project. These supports from MEXT and ICL
drilling at Nikawa revealed that shear failure could possi- are properly appreciated. Thanks also go to all colleagues
bly occur along the boundary between a granitic gravel of the Research Centre on Landslides of the DPRI of Kyoto
stratum and a layer of sand. The results of an undrained University for their cooperation on this study.
ring shear test on a sample taken from the granitic sand
layer were presented in this paper. The results showed
that sliding surface liquefaction could be triggered by an References
earthquake similar to the 1995 Kobe earthquake.
Application of the ring shear test to soils from the Alarcon-Guzman A, Leonards GA, Chameau JL (1988) Undrained
monotonic and cyclic strength of sands. J Geotech Eng-Asce
Minamata debris flow simulated a landslide triggered by 114(10):10891109
the gradual increase of ground-water level during heavy Ambraseys NN, Bommer JJ (1991) The attenuation of ground accel-
rainfall. Sliding surface liquefaction resulted with the erations in Europe. Earthquake Eng Struc 20:1791202
progress of shearing even in the naturally drained con- Bishop AW, Green GE, Garga VK, Andersen A, Brown JD (1971) A
dition corresponding to the gradual increase of pore new ring shear apparatus and its application to the measurement
of residual strength. Gotechnique 21(1):273328
water pressure in the debris-flow materials. An undrained Bromhead EN (1979) A simple ring shear apparatus. Ground Eng
ring shear test was carried out simulating the undrained 12(5):4044
loading process that takes place in the pre-existing tor- Bromhead EN (1986) The stability of slopes. Surrey University Press,
rent deposits and the weathered surface soils on the val- London, 373 p
ley slopes. This revealed that only a small increment in Castro G (1969) Liquefaction of sands. Ph.D. Thesis, Harvard Uni-
versity, Cambridge, Massachusetts
the shear stress due to the impacting by the displaced Castro G, Enos JL, France JW, Poulos SJ (1982) Liquefaction induced
soil mass could have caused shear failure of these depos- by cyclic loading. Report to National Science Foundation, Wash-
its/soils with the generation of high pore water pressure. ington, DC, No. NSF/CEE-82018
98 Kyoji Sassa Hiroshi Fukuoka Gonghui Wang Fawu Wang

Fukuoka H, Sassa K, Wang G, Sasaki R (2006) Observation of shear Sassa K (2004) Undrained dynamic-loading ring-shear appara-
zone development in ring-shear apparatus with a transparent tus and its application to landslide dynamics. Landslides 1(1):
shear box. Landslides 3(3):239251 719
Fukuoka H, Sassa K, Wang G (2007) Influence of shear speed and nor- Sassa K, Fukuoka H, Scarascia-Mugnozza G, Evans S (1996) Earth-
mal stress on the shear behavior and shear zone structure of granu- quake-induced-landslides: distribution, motion and mechanisms.
lar materials in naturally drained ring shear tests. Landslides (in Special issue for the great Hanshin Earthquake Disasters, Soils
print) published online, DOI:10.1007/s1034600600530 and Foundations 5364
Fukushima Y, Tanaka T (1990) A new attenuation relation for peak Sassa K, Fukuoka H, Wang FW (1997) Mechanism and risk assess-
horizontal acceleration of strong earthquake ground motion in ment of landslide-triggered-debris flows: lesson from the
Japan. Bull Seismol Soc Am 80(4):75783 1996.12.6 Otari debris flow disaster, Nagano, Japan. In: Cruden
Garga VK, Sendano JI (2002) Steady state strength of sands in a con- DM, Fell R (eds) Landslide risk assessment. Proceedings of the
stant volume ring shear apparatus. Geotech Test J 25(4):414421 International workshop on landslide risk assessment. Honolulu,
Gibo S (1994) Ring shear apparatus for measuring residual strengths 1921 February, pp 347356
and its measurement accuracy. Journal of Japan Landslide Soci- Sassa K, Fukuoka H, Wang FW (1998) Mechanism of rapid long
ety 31(3):2430 run-out motion in the Sumikawa reactivated landslide in Akita
Hungr O, Morgenstern NR (1984) High velocity ring shear tests on Prefecture and the Harihara landslidedebris flow in Kago-
sand. Gotechnique 34(3):415421 shima Prefecture, 1997, Japan. J Jpn Landslide Soc 35(2):2937
Hvorslev MJ (1939) Torsion shear tests and their place in the deter- (in Japanese)
mination of the shearing resistance of soils. Proc Am Soc Test Sassa K, Wang FW, Wang GH (1999) Mechanism of rapid landslides
Mater 39:9991022 by the ring shear tests on the long run-out rapid landslide in
Ishihara K (1993) Liquefaction and flow failure during earthquakes. Hiegaeshi, Saigo Village, Fukushima Prefecture. In: Proceedings
Gotechnique 43(3):349451 Symposium on1998 Slope Disasters and Sediment Disasters
Kramer KL, Seed HB (1988) Initiation of soil liquefaction under static Characteristics and Actual Conditions, Japan Landslide Society,
loading conditions. J Geotech Eng-ASCE 114:412430 pp 3849 (in Japanese)
Okada Y, Sassa K, Fukuoka H (2000) Liquefaction and the steady Sassa K, Wang G, Fukuoka H (2003a) Assessment of earth-
state of weathered granitic sands obtained by undrained ring quake-induced catastrophic landslides in urban areas and
shear tests: a fundamental study on the mechanism of liquidized their prevention planning. In: Proceedings of International Con-
landslides. J Natural Disaster Sci 22(2):7585 ference on Slope Engineering, December 2003, Hong Kong, 1,
Poulos SJ (1981) The steady state of deformation. J Geotech Eng- pp 2649
ASCE 107 (GT5):553562 Sassa K, Wang G, Fukuoka H (2003b) Performing undrained shear
Poulos SJ, Castro G, France JW (1985) Liquefaction evaluation pro- tests on saturated sands in a new intelligent type of ring shear
cedure. J Geotech Eng-ASCE 111(6):772792 apparatus. Geotech Test J 26(3):257265
Rouse WC (1984) Flowslides. In: Brunsden D, Prior DB (eds) Slope Savage SB, Sayed M (1984) Stresses developed in dry cohesionless
instability. John Wiley & Son, New York, pp 491522 granular materials sheared in an annular shear cell. J Fluid Mech
Sassa K (1984) The mechanism starting liquefied landslides and 142:391430
debris flows. In: Proceedings of 4th International Symposium on Shimamoto T, Tsutsumi A (1994) A new rotary-shear high speed fric-
Landslides, Toronto, June, 2, pp 349354 tional testing machine: its basic design and scope of research.
Sassa K (1988) Geotechnical model for the motion of landslides. In: Structural Geology 39:6578 (in Japanese)
Special Lecture of 5th International Symposium on Landslides, Shoaei Z, Sassa K (1994) Basic study on the shear behavior of land-
Landslides, 1015 July, 1, pp 3755 slides during earthquakes excess pore pressure in the undrained
Sassa K (1994) Development of a new cyclic loading ring shear appa- cyclic loading ring shear tests. Bull Disaster Prev Res Inst Kyoto
ratus to study earthquake-induced-landslides. Report for Grain- Univ 44(1):143
in-Aid for Developmental Scientific Research by the Ministry of Sladen JA, DHollander, RD, Krahn J (1985) The liquefaction of sands,
Education, Science and Culture, Japan (Project No. 03556021), 106 p a collapse surface approach. Can Geotech J 22:564578
Sassa K (1995) Access to the dynamics of landslides during earthquakes Tika TM (1989) The effect of rate of shear on the residual strength
by a new cyclic loading high-speed ring shear apparatus (keynote of soil. PhD thesis, University of London (Imperial College of
paper). In: 6th International Symposium on Landslides,Landslides, Science and Technology), 494 p
A.A. Balkema. Christchurch, 1014 February, 3, pp 19191937 Tika TE, Hutchinson JN (1999) Ring shear tests on soil from the Vaiont
Sassa K (1996) Prediction of earthquake induced landslides. In: landslide slip surface. Gotechnique 49(1):5974
Proceedings of 7th International Symposium on Landslides, A.A. Vaid YP, Chern JC (1985) Cyclic and monotonic undrained response
Balkema, Trondheim, 1721 June, 1, 115132 of saturated sands. Khosla V (ed) Advances in the art of testing
Sassa K (1997) A new intelligent-type dynamic loading ring shear soils under cyclic conditions. ASCE, pp 120147
apparatus. Landslide News 10:33 Vankov D, Sassa K (1999) Mechanism of earthquake-induced land-
Sassa K (1998) Recent urban landslide disasters in Japan and their mech- slides on almost flat slopes studied with a ring shear apparatus.
anisms. In: Proceedings 2nd International Conference on Environ- J Natural Disaster Sci 21(1):2335
mental Management, Environmental Management, 1, pp 4758 Voight, B (ed) (1978) Rockslides and avalanches 1: natural phenom-
Sassa K (2000) Mechanism of flows in granular soils. In: Proceed- ena. Elsevier, Amsterdam
ings of the International Conference of Geotechnical and Geo- Voight, B (ed) (1979) Rockslides and avalanches 2: engineering sites.
logical Engineering, GEOENG2000, Melbourne, 1, pp 16711702 Elsevier, Amsterdam
Sassa K (2002) Study on the mechanism of earthquake and rainfall Wang G, Sassa K, Fukuoka H (2003) Downslope volume enlargement
induced rapid flow phenomena- Disasters and their mitigation in of a debris slide-debris flow in the 1999 Hiroshima, Japan, rain-
large-scale cities. In: Proceedings of the Symposium on Aerial Pre- storm. Eng Geol 69:309330
diction of Earthquake and Rainfall Induced Flow Phenomena Zhang S, Tullis TE, Scruggs VJ (1999) Permeability anisotropy and
(APERIF), New century of urban area landslide disaster mitiga- pressure dependency of permeability in experimentally sheared
tion, Tokyo. 31 August 1 September 2002, pp 733 (in Japanese) gouge materials. J Struct Geol 21(7):795806
Chapter 7

Shear Behavior and Shear Zone Structure of Granular Materials

in Naturally Drained Ring Shear Tests

Hiroshi Fukuoka* Kyoji Sassa Gonghui Wang

Abstract. Sliding Surface Liquefaction is a process causing strength under saturated undrained conditions. They named the
loss and consequent rapid motion and long runout of certain land- mechanism as Sliding Surface Liquefaction, (Sassa 1996,
slides. Using a new ring shear apparatus with a transparent shear-
2000; Sassa et al. 1996) hereafter often abbreviated as SSL.
box and digital video camera system, shear-speed-controlled tests
were conducted on mixed grains (mixture of three different sizes of Studies of the friction of granular materials for the
sand and gravel) and mixed beads to study shear behavior and shear purpose of landslide runout mechanism were conducted
zone development process under the naturally drained condition by many researchers using geological, geotechnical, and
in which pore pressure is allowed to dissipate through the opened geophysical approaches. The effect of shear speed and
upper drainage valve during shearing. Higher excess pore water
normal stress on shear characteristics of granular mate-
pressure and lower minimum apparent friction were observed in
the tests where grain crushing was more extensive under higher rials had been widely examined (Novosad 1964; Scarlett and
normal stress and higher shear speed. Along with the diffusion of Todd 1969; Bridgwater 1972; Hungr and Morgenstern 1984;
silty water generated by grain crushing, smaller particles were trans- Sassa 1984; Vibert et al. 1989; Fukuoka et al. 1990; Fukuoka
ported upward and downward from the shear zone. Concentration 1991; Tika 1989; Tika and Hutchinson 1999; Lemos 2003).
of larger grains to the central and upper part of the shear zone was These experimental studies were conducted by ring shear
confirmed by means of visual observation together with grain size
analysis of sliced samples from several layers after the test. On the
apparatus or torsion shear apparatus. Existing appara-
other hand, smaller particles were accumulated mostly below the tuses have not allowed the experimenters to observe the
layer where larger grains were accumulated. The reason why larger grains of the sample during the test, nor to detect the de-
grains were accumulated into the shear zone may be interpreted velopment of a shear zone during shearing. Lang et al.
as follows: grains under shearing are also subjected to vertical move- 1991 tried to use a transparent shear box for the low-stress
ment, the penetration resistance of larger grains into a layer of mov-
ring shear apparatus which was developed by Sassa 1984
ing particles is smaller than that into the static layer. Therefore, larger
grains tend to move into the layer of moving grains. At the same and used a video image processing system to track the
time, smaller particles can drop into the pores of underlying larger movement of the samples grains. They conducted a se-
grains downward due to gravity. ries of tests on glass beads of 26 mm diameters under
low normal stresses of 2.9 kPa and 29 kPa and shear
Keywords. Naturally drained and speed-controlled ring shear tests,
speeds of 1 cm s1 and 10 cm s1. They found that shear
transparent shear box, shear zone development, concentration of
grains and particles, velocity distribution profile, grain size distribution zone thickness decreased under higher normal stress in
the test on beads of smaller diameter, and that the shear
zone thickness was slightly greater during faster shear.
However, these tests used only glass beads and did not
7.1 Introduction use natural materials. Moreover, their apparatus was not
capable of maintaining undrained conditions, or to mea-
Every year, many rapid and long-travel landslides take sure pore water pressure.
place due to earthquakes and heavy rainfall. These land- Sassa and his colleagues have developed seven designs
slides brought about casualties and property damage, of ring shear apparatus since 1984 (Sassa et al. 2004a).
worsened by the widespread development of towns and The first apparatus (DPRI Ver.1) was used in the study of
cities into upland areas. To reduce such landslide disas- Vibert et al. (1989), Lang et al. (1991), and DPRI Ver.2 was
ters, the development of reliable risk assessment has the used by Fukuoka (1991). Wang and Sassa 2002 examined
first priority. Practical assessment of future landslide the shear zone structure by exposing sections of silica
runout distance towards development areas requires pre- sand samples in the ring shear apparatus DPRI Ver.6 af-
cise prediction of the friction characteristics of the mate- ter undrained tests of large shear displacement. However,
rials. Based on geotechnical and experimental studies their sample box of DPRI Ver.6 was metallic and they could
using an undrained ring shear apparatus on the January not observe the sample during shear. Wafid et al. (2004)
1995 Nikawa landslide in Japan, Sassa and colleagues investigated the development of shear zones in undrained
found that granular materials show very high mobility ring shear tests by observing the sections of sand samples
100 Hiroshi Fukuoka Kyoji Sassa Gonghui Wang

stopped at different stages of shear displacement from the

7.2 Ring Shear Apparatus and Observation System
initiation of failure to the steady state. They found coarse
grains accumulated in the shear zone and proposed a seg-
regation model for shear zone development under The ring shear apparatus DPRI Ver.7, the latest model of
undrained conditions. However, continuous monitoring undrained ring shear apparatus, has a transparent shear
from outside of the shear box and detection of horizontal box which enables the observation of sand grains during slow
and vertical movement of the grains were still impossible. to high speed shearing (Sassa et al. 2004a). The overview of
The authors deployed the DPRI Ver.7 with a transpar- the ring shear apparatus DPRI Ver.7 is shown in Fig. 7.1 and
ent shear box and conducted shear-speed-controlled tests its transparent shear box is shown in Fig. 7.2. The basic
on coarse grained silica sands to study the shear zone for- structure of this apparatus is the same as DPRI-5 and
mation process in granular materials (Fukuoka et al. DPRI-6 which are introduced by Sassa et al. (2003, 2004).
2005a,b). Velocity distribution profiles of grains under The basic structure of the DPRI Ver.7 is similar to its
shear at various stages in the ring shear tests were ob- predecessors DPRI Ver.5 and DPRI Ver.6, introduced by
served through processing the video image. They found Sassa et al. (2003, 2004a). The specifications of the DPRI
that after shear resistance reached peak strength, the Ver.7 apparatus are given by Sassa et al. (2004a) and
thickness of the shear zone tends to decrease. However, Fukuoka et al. (2005a,b). The most important detail of
these two studies focused on the variation of the velocity the DPRI Ver.7 is the transparent sample box which al-
distribution profiles, and the variation of mechanical lows direct observation of grains inside the shear box
properties was not examined in thorough detail. during shearing from outside.
In this study, the main purpose is to study the influ- The schematic illustration of the samples in the appa-
ence of shear speed and normal stress on the shear be- ratus and set-up of two digital video cameras are shown
havior from an approach of shear zone development pro- in Fig. 7.3a. Movies of the sample during shear are taken
cess observation through examining mechanical proper- by two digital video cameras (hereafter, abbreviated as
ties and variation of grain size distribution in the sample. DV cameras) 1 and 2. Initially, only a floor-mounted DV
camera 1 was used. The image obtained by DV camera 1
is shown in the middle photo of Fig. 7.3b. As shown in
the left photo of Fig. 7.3b, the image of the upper and lower
shear box sample is clear enough to distinguish each sand
grain before shear. However, during shear, the image of
the lower shear box (inside the blue box in the middle
photo of Fig. 7.3b) was severely blurred and it became
impossible to distinguish each grain. Another digital
video camera (fixed on the rotating table of the lower shear
box) shown as DV camera 2 in Fig. 7.3a was then in-
stalled. In the right photo of the Fig. 7.3b is a sample im-
age taken by DV camera 2 during shear. Although grains
in the upper shear box (in the blue box) cannot be distin-
guished in this photo because the upper shear box was

Fig. 7.1. Overview of the ring shear apparatus with a transparent

shear box (DPRI Ver.7) Fig. 7.2. Transparent shear box of DPRI Ver.7
Chapter 7 Shear Behavior and Shear Zone Structure of Granular Materials in Naturally Drained Ring Shear Tests 101

rotating during shear, the image of the grains below the

7.3 Samples and Their Physical Properties
shear plane is clear enough for them to be distinguished.
A 2.0-mm-thick stainless steel plate is attached to the
bottom of the upper half of the outer shear box contact- In this study, two types of samples were used: the first sample
ing the rubber edge. This plate is necessary to protect the is mixed sands consisting of gravels and silica sands No. 1
acrylic shear box from damage due to heating during high (S1) and No. 4 (S4). The second sample is mixed beads. The
speed shearing. This system did not allow the observa- properties of the samples are summarized in Table 7.1.
tion of the exact shear plane and its vicinity, because the Gravel used in this study consists of not completely round
shear zone is shaded by invisible contacting parts con- grains, but with smooth surfaces. Basic properties of the grav-
sisting of the rubber edges and the metal plate shown as els are as follows: mean diameter D50 = 6.5 mm, uniformity
the invisible part in Fig. 7.3b. Pore pressure transduc- coefficient Uc = 1.47, and specific gravity Gs = 2.65. Silica
ers were installed in the upper and inner sample box sand is produced artificially by crushing of natural rocks
(Sassa et al. 2004a; Fukuoka et al. 2005b). and sieving. It is commonly used as construction material.
This silica sand consists of roughly round grains, of which
9298% are quartz and a little amount of feldspar. Basic
properties of silica sands S1 and S4 are as follows: silica
sand S1 has a mean diameter of D50 = 3.01 mm, uniformity
coefficient of Uc = 1.64, and specific gravity of Gs = 2.64.
Silica sand S4 has D50 = 1.05 mm, Uc = 2.35, and Gs = 2.64.
Mixed beads consist of glass beads, plastic beads, and
polystyrene beads. The basic properties of glass beads are
D50 = 2.0 mm, Uc = 1.0 (single-size grains), and Gs = 2.52;
plastic beads are D50 = 6.0 mm, Uc = 1.0, and Gs = 1.77;
polystyrene beads are D50=1.0 mm, Uc = 1.0, and Gs = 1.00.
Each of the three types of beads is mixed by equal per-
centage in bulk volume.
In this study, the tests were performed on the mixed sands
and mixed beads, under different normal stresses and shear
speeds, then the dynamic behaviours were observed.

7.4 Testing Conditions and Procedure

The test series consists of three ring shear tests on mixed

sands and one test on mixed beads. The conditions of the
four tests are summarized in Table 7.2. In the all tests, the
speed-controlled mode is employed to examine shear
speed dependency of shear resistance. In the initial stage
of each test, shear speed was increased at a constant ac-
Fig. 7.3. a Schematic illustration of the samples in the ring shear ap- celeration rate until shear speed reached the target speed.
paratus DPRI Ver.7 and the digital video cameras 1 and 2 for video Stress and shear speed conditions of tests S1, S2, and S3
shooting; b shooting area of the digital video (DV) cameras 1 and 2.
The image of the lower shear box taken by DV camera 1 is blurred.
on mixed silica sand were selected from combinations of
DV camera 2 installed on the rotating table of the lower shear box high normal stress (200 kPa) and low normal stress (22 kPa)
can take clear video images of the samples conditions and high shear speed (150 cm s1) and low
102 Hiroshi Fukuoka Kyoji Sassa Gonghui Wang

shear speed (9 cm s1) conditions. Test S1 was conducted

7.5 Test Results on Mixed Sands
under low normal stress and high shear speed condition,
test S2 was under high stress and high shear speed condi-
tions, and test S3 was under high stress and low shear speed 7.5.1 Video Observation of Shear Zone Development
conditions. Test B1 was conducted on mixed beads under with Progress of Shear Displacement
low normal stress and high shear speed.
In this study, all of the tests were conducted under natu- Figures 7.4, 7.5 and 7.6 present series of photos of the
rally drained condition keeping the valve of the drainage samples within the shear box at four different shear dis-
in the upper sample box open during shear. In this con- placements, i.e., 0, 30, 300, and 3000 cm of the three tests
dition, pore pressure during shear is not always zero. In of S1, S2, S3, respectively. Photos at each shear displace-
the actual landslides, shearing along the sliding surface is ment are combinations of the images of the upper and
neither completely undrained condition (no pore pres- lower shear boxes taken by the two DV cameras. In the
sure dissipation), nor completely drained condition (no study by Sassa et al. (2004a), the series of photos of the
excess pore pressure) as discussed by Sassa et al. (2004b) test on the silica sand No. 1 under high speed shear
where the effects were examined for the Nikawa landslide (200 cm s1) and higher normal stress (200 kPa) by the
triggered by the Hyogoken-Nambu earthquake. Pore pres- DPRI Ver.7 showed obvious silty water generation during
sure build-up is a combined effect of the pore pressure shear. They explained that when grain crushing created
generation rate in the shear zone and pore pressure dissi- small grains, pore water became silty and gradually dif-
pation rate which are functions of shear speed, permeabil- fused later. In the study by Fukuoka et al. (2005a,b), silty
ity of soil layers, drainage path, etc. The purpose of the ring water was not obvious in their tests on silica sand No. 1
shear tests is to mechanically simulate shearing in real land- under lower normal stress (22 kPa), due to the fact that
slides. In real landslides, upward drainage is allowed. There- the normal stress of the tests was too low to cause grain
fore, the shearing was conducted keeping the upper drain- crushing for making the samples silty. In Fig. 7.4, silty
age valve open allowing the pore pressure dissipation only water was generated and diffused, but some grains still
upward. This test procedure may not reproduce the natural visible at a shear displacement of 3000 cm. On the other
condition in a quantitative way, but it may illuminate the hand, in the series of photos in Figs. 7.5 and 7.6, silty wa-
pore pressure build-up behavior in a qualitative manner. ter diffused so much in both the upper and lower shear
The testing procedure was as follows: (1) samples were boxes that original grains could not be visible at a shear
placed into the sample box by dry deposition and CO2 gas displacement of 3000 cm. The difference between test S1
was supplied to expel the air. Then, de-aired water was in- and tests S2 and S3 is mainly due to the different normal
filtrated from the bottom slowly under a pressure of about stresses, i.e., higher normal stress in tests S2 and S3 pro-
1 m head to saturate the sample; (2) a targeted normal stress motes more intensive grain crushing and creates more
was loaded slowly under the naturally drained conditions, amounts from smaller grains to make the pore water silty.
keeping the upper drainage valve open; (3) the sample was
then sheared under the shear-speed-controlled condition
and the naturally drained condition as shown in Table 7.2 7.5.2 Variation of Stress and Pore Pressure with
until the shear speed reached the target value; (4) shearing Progress of Shear Displacement
was continued at constant shear speed until the shear dis-
placement reached 3000 cm (30 m). During shearing, DV Variation of the shear resistance, pore pressure, and
cameras 1 and 2 were monitoring the upper and lower shear sample height increment against the shear displacement
box, respectively; (5) after the termination of shearing, the of tests S1, S2, and S3 are shown in Figs. 7.7, 7.8 and 7.9.
upper sample box was dismounted to expose the sample; In these tests, the sample height continued reducing from
(6) samples were taken from aeach layer of about 1 cm thick- the beginning to the end. In test S1 (shear speed: 150 cm s1,
ness from top to bottom of the samples within the sample normal stress: 22 kPa) (Fig. 7.7), the sample height reduc-
box; finally, (7) the grain size distribution of each layer was tion at shear displacement of 3000 cm was about 1.1 mm.
measured by sieving. On the other hand, tests S2 and S3 under high normal
Chapter 7 Shear Behavior and Shear Zone Structure of Granular Materials in Naturally Drained Ring Shear Tests 103

Fig. 7.4.
A series of photos of the upper
and lower shear box of the
ring shear test S1 on mixed
silica sands under the naturally
drained condition. Normal stress
= 22 kPa, shear speed= 150 cm s1,
acceleration at the initial stage
was 50 cm s2. Video images of
the upper shear box were taken
by DV camera 1 which was lo-
cated on the laboratory floor,
and the images of lower shear
box were taken by DV camera 2
which was fixed on the rotating
table of the lower shear box

Fig. 7.5.
A series of photos of the upper
and lower shear box of the
ring shear test S2 on mixed
silica sands under the naturally
drained condition. Normal stress
= 200 kPa, shear speed= 150 cm s1,
acceleration = 50 cm s2
104 Hiroshi Fukuoka Kyoji Sassa Gonghui Wang

Fig. 7.6.
A series of photos of the upper
and lower shear box of the
ring shear test S3 on mixed
silica sands under the naturally
drained condition. Normal stress
= 200 kPa, shear speed = 9 cm s1,
acceleration = 2 cm s2

erated after shear resistance reached peak strength, i.e.,

after the shear failure, in all of the tests of S1, S2, and S3.
The shear resistance dropped due to the generated excess
pore pressure and then increased again along with the
dissipation of the excess pore pressure.

7.5.3 Concentration of Larger Grains in the Shear Zone

during Shearing and Resulting Variation of Grain
Size Distribution

In Figs. 7.4, 7.5 and 7.6, it can be seen that the proportion
of smaller particles was increased with shearing, due to
Fig. 7.7. Relation between shear displacement vs shear resistance, pore
the grain crushing in the shear zone. To compare the re-
pressure, and sample height increment in test S1 (150 cm s1, 22 kPa) sults of the crushable sands with the uncrushable samples,
on mixed silica sands in the naturally drained condition another test B1 was performed on the mixed beads, where
the test conditions were the same as those in test S1.
stress (200 kPa) resulted in much greater sample height The series of photos at four stages from before shear-
reduction (about 14 mm in the test S2, and about 18 mm ing until the end of shearing taken in test B1 are presented
in the test S3). This difference of sample height reduc- in Fig. 7.8. White beads are 6.0 mm in diameter, blue beads
tion between high and low normal stresses can be ex- are 1.0 mm, and transparent beads (not so visible in this
plained by different grain crushing degrees at different figure) are 2.0 mm. In this figure, before shearing, white
levels of normal stress. and blue beads are distributed almost uniformly. How-
Because all of the tests were conducted under the natu- ever, a blue zone appeared at shear displacement of
rally drained condition, pore pressure could change dur- L = 30 cm just above the shear zone in the upper shear
ing the test. It is noted that excess pore pressure was gen- box. Thereafter, the blue band of small beads appeared
Chapter 7 Shear Behavior and Shear Zone Structure of Granular Materials in Naturally Drained Ring Shear Tests 105

Fig. 7.8. Relation between shear displacement vs shear resistance, pore Fig. 7.9. Relation between shear displacement vs shear resistance,
pressure, and sample height increment in test S2 (150 cm s1, 200 kPa) pore pressure, and sample height increment in test S3 (9 cm s1,
in the naturally drained condition 200 kPa)

Fig. 7.10.
Series of photos in the ring shear
test B1 on mixed sands under low
normal stress (22 kPa) and high
shear speed (150 cm s1) in the
naturally drained condition. L is
the shear displacement at each
photo. In zones A and C, fine
blue beads are concentrated,
while white coarser beads are
concentrated in the shear zone B.
The concentration of coarse
grains in the shear zone was also
confirmed in the vertical section
after the test

also in the lower shear box as shown in Fig. 7.10 at the Figures 7.11, 7.12 and 7.13 show the results of grain
shear displacement of L = 300 cm and 3000 cm, which are size analysis performed for sand samples from different
marked as A and C. On the other hand, comparing the layers for tests S1, S2, and S3, respectively. On the left of
images of the shear zone at L = 300 and 3000 cm, the num- each figure, the grain size distribution of each sliced part,
ber of greater white beads seemed to increase in the which is marked as AF or AG, is presented in color bars.
3000 shear displacement, forming a layer, which is marked The corresponding photo of a vertical section view of the
as B in the photo, while small blue beads concentrated in sample taken after ring shear test is shown on the right of
the two layers of A and C, located just above and below each figure. The red dotted lines in these photos show
the layer B. the location of the shear plane, i.e., the boundary between
106 Hiroshi Fukuoka Kyoji Sassa Gonghui Wang

Fig. 7.11.
Left: Grain size distribution from
thin layers of the sample in the
naturally drained ring shear
test S1 on mixed sands under
low normal stress (22 kPa) and
high shear speed (150 cm s1).
Right: Photo of the vertical sec-
tion of the sample after the ring
shear test. The layer marked as
AF corresponds to the same
mark on the left figure

Fig. 7.12.
Left: Grain size distribution from
thin layers of the sample in the
naturally drained ring shear
test S2 (200 kPa, 150 cm s1) on
mixed sands. Right: Photo of the
vertical section of the sample
after the ring shear test. The
layer marked as AF corresponds
to the same mark on the left

Fig. 7.13.
Left: Grain size distribution from
thin layers of the sample in the
naturally drained ring shear
test S3 (200 kPa, 9 cm s1) on
mixed sands. Right: Photo of the
vertical section of the sample
after the ring shear test. The
layer marked as AF corresponds
to the same mark on the left

the upper shear box and the lower shear box of the appa- 0.85 mm as S-particles (blue bars), and smaller than
ratus. Hereinafter, grains and particles of which the size 0.85 mm as SS-particles (yellow bars). The color bar
is between 9.5 mm and 4.75 mm are denoted as LL-grains marked as Original presents the grain size distribution
(percentage in weight is presented as green bars), 4.75 mm of the original samples before the ring shear test and the
and 2.0 mm as L-grains (purple bars), 2.0 mm and one at the bottom of the figure marked as After shear-
Chapter 7 Shear Behavior and Shear Zone Structure of Granular Materials in Naturally Drained Ring Shear Tests 107

ing presents the average grain size distribution of the ated at the same height area. Therefore, it was estimated
samples after the ring shear test, which were collected that fine particles generated within the shear zone moved
from top to bottom within the samples. upward along with the diffusion of silty water due to high
Figure 7.11 shows the grain size distribution of the excess pore water pressure in the shear zone.
sample in the test S1. In the right photo, LL-grains of Figure 7.13 for high normal stress (200 kPa) and low
darker color concentrated above the shear plane. The per- shear speed (9 cm s1) test shows the grain size distribution
centage of LL-grains in the color bars of C and D is much and a photo of vertical section of the sample after the test S3.
larger than the original. Especially LL-grains of D is al- In the photo, the concentration of LL-grains was not so vis-
most double the original. L-grains concentrated in the ible as Fig. 7.11. When the increase of SS-grains amount is
bars of E and F above the LL-grain concentrating zone. compared to the differences of Original and After shear-
S-particles concentrated in the two bars A and B located ing with those in Figs. 7.11, 7.12 and 7.13, obviously the
below the LL-grain dominant zone. It was observed that greatest amount of new SS-particles was generated by
S-particles and SS-particles settled downward under this grain crushing in Fig. 7.13 for test S3. The difference of
low normal stress test (S1) immediately after shearing was conditions between tests S2 (Fig. 7.12) and S3 (Fig. 7.13) was
stopped at shear displacement of 3000 cm. Therefore, the only shear speed. S3 (9 cm s1) took 15 times longer time to
amount of S-particles and SS-particles in the bars of D, E, reach 30 m shearing displacement than S2 (150 cm s1).
and F are smaller than the original. During shear, the con- Therefore, S-particles generation will be not only the func-
tents of S-particles and SS-particles must be a little more tion of shear displacement and normal stress, but also
than that presented in the upper shear box and a little affected by shearing time.
less in the lower shear box. The total amount of new SS- The SS-particles concentrated around the shear plane
particles generated during shear by grain crushing is pre- and below the shear plane in the lower shear box as shown
sented by the difference between the colored bars of in the four colored bars of A, B and C in Figs. 7.11 and
Original and After shearing and it is only about 4%. 7.12 and A, B, C and D in Fig. 7.13. On the other hand, the
Consequently, the S-particles and SS-particles concen- LL-grains concentrated in the two colored bars of C and
trated into the lower box, and the LL-grains and L-grains D in Figs. 7.11 and 7.12. However the observed amount
concentrated around and above the shear plane. of the LL-grains of the two colored bars of C and D was
Figure 7.12 shows the results for test S2. Because the almost same as the Original sample in Fig. 7.13. The
normal stress of test S2 is much higher than test S1 in amount of LL-grains of the three colored bars of A, B and
Fig. 7.11, more SS-particles were generated by grain crush- E, located above and below the slices of C and D in the
ing as shown by the difference of SS-particles content vertical section, was smaller than C and D in Figs. 7.11,
between the colored bars of Original and After shear- 7.12 and 7.13. This result suggests a possibility that there
ing. In the photos of Fig. 7.12 for the higher normal stress is a mechanism which lets the LL-grains concentrate
tests (200 kPa), the concentration of LL-grains in the vi- around the shear plane.
cinity of the shear plane was not so visible. SS-particles,
most of which may be generated by grain crushing, are
distributed throughout the vertical section. The grain size 7.6 Discussion
distribution in Fig. 7.12 shows that SS-particles concen-
trated in the three colored bars of A, B, and C, implying 7.6.1 Influence of Shear Speed and Normal Stress on
occurrence of deposition toward the bottom of the shear the Mobilized Shear Resistance
box. This figure also shows that LL-grains concentrated
around the shear surface of the two bars of C and D. How- In Fig. 7.14a, variations of apparent friction and pore
ever, their percentage was not so high as the bars of C and pressure ratio of the mixed sands tests S1, S2, and S3 are
D in Fig. 7.11 for lower normal stress test (22 kPa). The shown against the shear displacement from 0.1 cm to
reason for the fact that accumulation of LL-grains oc- 3 000 cm. In Fig. 7.14b, the same parameters were plot-
curred extensively under lower normal stress must be the ted vs elapsed time. Shear displacement and elapsed time
only difference in their test condition of the normal stress are presented in log scale. Apparent friction, which is
between test S1 (22 kPa) and test S2 (200 kPa). The con- represented by red lines, is calculated by mobilized shear
tents of S-particles and SS-particles within the top two resistance divided by total normal stress . Apparent
colored bars of F and G in Fig. 7.12 were even greater than friction defined as tan1( /) by Sassa (1988) is used as
the original sample. Considering that in Fig. 7.5 sand the most important index for landslide mobility. When
grains near the top of the sample beneath the loading plate, apparent friction is reduced in the shear zone, higher
corresponding to the two colored bars of F and G in mobility is expected.
Fig. 7.12 showed that most of the grain did not move up Pore pressure ratio ru, which is represented by blue
to 300 cm of shear displacement, and there is a possibil- lines, is calculated by generated excess pore pressure u
ity that the new S-particles and SS-particles were not cre- divided by total normal stress . Values of ru are not
108 Hiroshi Fukuoka Kyoji Sassa Gonghui Wang

The effect of shear speed is obvious by comparing the

results of tests S2 and S3, where the applied normal stress
were 200 kPa, but the shear speed was 150 cm s1 for S2,
and 9 cm s1 for S3, respectively. The peak excess pore
pressure ratio of S2 is about 0.65, and that of S3 is about
0.16. The peak excess pore pressure of test S2 was almost
four times that in test S3. The large difference of the peak
excess pore pressure may be caused by the large differ-
ence in the shear speed. The difference in the build-up
rate of excess pore pressure is about 15 times between S2
and S3. However, the dissipation rate is almost the same
among tests S1, S2, and S3, because dissipation depends
on pore pressure difference, dissipation path length and
permeability. The values of these factors are almost the
same (except the first factor). Thus, the peak excess pore
pressure which was built up under the naturally drained
condition is higher in the tests under higher shear speed.
Thus, it is reasonable that the observed peak excess pore
pressure in test S2 was much greater than S1. Conse-
Fig. 7.14. Variation of apparent friction, pore pressure ratio vs
quently, in the naturally drained ring shear tests under
(a) shear displacement, (b) elapsed time in log scale in the naturally higher shear speed and higher normal stress condition,
drained test of mixed sands higher excess pore pressure and lower minimum appar-
ent friction can be expected.
greater than 1.0. When the sample nears a state of lique- On the other hand, when the normal stress condition
faction or sliding liquefaction, ru approaches 1.0. As ex- is compared, S1 was under 22 kPa and S2 was under
plained in previous sections, the ring shear tests of this 200 kPa, and both tests were conducted under the
study were conducted under the naturally drained con- same shear speed of 150 cm and the same acceleration
dition. Excess pore pressure was generated during shear- of 50 cm s2. The peak excess pore pressure in test S2
ing of each test, but was naturally dissipated through the is about 0.65 and that in test S1 is about 0.31. The peak
upper drainage line of the shear box. excess pore pressure of test S2 showed about two times
In Fig. 7.14a, apparent friction once increased as the of that in test S1. The final sample height reduction in
shear resistance reached the peak strength in red plots of test S1 was 1.1 mm (Fig. 7.7), but that in test S2 was
S1 and S2. The plots of S3 reached peak before shear dis- 14 mm (Fig. 7.8), more than ten times the value in S1.
placement of about 0.7 cm. But after reaching the peak Therefore, the difference of pore pressure build-up
strength, shear resistance of each test turned to be reduced was caused by the difference of grain crushing and the
and showed a minimum value as the excess pore pressure resulting pore pressure generation due to a difference of
was generated. The characteristics of this figure were: normal stress.
(1) order of the value of the peak ru is S2 > S1 > S3; (2) or-
der of the local minimum value of apparent friction in
each test is S2 < S1 < S3. In general, when the excess pore 7.6.2 Influence of Normal Stress and Shear Speed on
pressure ratio is higher, the apparent friction is lower. The Grain Crushing and Concentration of the Coarse
shear displacement at which the peak value of ru appears Grain in the Shear Zone
in each test is larger than the one at which the smallest
value of apparent friction appeared. In Fig. 7.15, the variation in the amount of LL-grains and
The drainage path of the apparatus is described in SS-particles in each layer of the sample is plotted. The
Sassa et al. (2004a). Pore pressure is measured at the an- horizontal axis is the height from the shear plane. The
nular gutter just 2 mm above the shear plane and filled location of the shear plane is represented by the vertical
with metal filters and felting cloth. Pore pressure trans- straight line at height = 0 cm. The positive side of the
ducers are installed at the exit of the pipes extending from height from shear plane corresponds to the upper shear
the annular gutter. Although the samples are sandy ma- box, while the negative side corresponds to the lower
terial and permeability is higher than clayey materials, shear box. Blue plots and lines represent the variation of
slight delay of measuring excess pore pressure in the shear SS-particles amount, and red plots and lines represent
plane may take place due to the 2 mm distance from shear variation of LL-grains amount for each test of S1, S2, and
plane in these naturally drained tests. S3. Variation of the grains proportions in each layer of
Chapter 7 Shear Behavior and Shear Zone Structure of Granular Materials in Naturally Drained Ring Shear Tests 109

zone under high speed tests. This tendency is consistent

with the results of test B1 shown in Fig. 7.10, but in test S3,
the maximum ratio of the LL-grains was almost 1 prob-
ably affected by greater grain crushing and the resulting
less LL-grains.
One possible interpretation why large grains were ac-
cumulated in the vicinity of the shear zone is as follows:
grains under shearing are subjected to vertical force due
to dilatancy. Small particles can enter into the pores be-
tween the large grains, however, large grains can move
vertically only by the penetration into a layer by moving
Fig. 7.15. Variation of the rate of sample height reduction (dh/dt) and other particles and making spaces. If a layer of moving
its acceleration (d2h/dt2) for the shear displacement from 0.1 cm to particles such as the shear zone will exist on one side and
3 000 cm in log scale. Variations in the amount of LL-grains and SS- a layer of static layer exist on another side, the penetra-
particles in each layer of the sample, represented by the ratio be- tion of large grains into the layer of moving particles will
tween before and after the tests of S1 (22 kPa, 150 cm s1), S2 (200 kPa,
150 cm s1), and S3 (200 kPa, 9 cm s1). Blue plots and lines repre-
be much easier, because there is no need to spend energy
sent variation of SS-particle amounts, and red plots and lines repre- to force moving other particles. This mechanism can be
sent variation of LL-grain amounts understood by a simple experiment that a finger can be
inserted into sands under shaking condition much easier
the sample is represented by the ratio between those be- than into sands under static condition. Melosh (1986)
fore and after the tests. When the ratio is greater than tested a series of such experiments in which sands were
1.0, it means that the amount of grains increased and excited by a 37 kHz acoustic field and an aluminum rod
when it is smaller than 1.0, it means that the amount de- was pushed into the sands under various strain rate to
creased. measure shear penetration resistance variation compared
As for the effect of normal stress on grain crushing, with the one under static condition. The test results showed
the order of the peak value of SS-particles distribution obvious reduction in the 37 kHz acoustic condition.
was S3 > S2 > S1. This means that higher normal stress Therefore, large grains tend to move into the layer of
in tests S2 and S3 causes generation of greater amount of moving particles, namely, large grains above the shear
SS-particles than the low normal stress in test S1. On the zone tend to move downward to the shear zone, and large
other hand, higher shear speed in test S2 causes less amount grains below the shear zone tend to move upward to the
of SS-particles than the low shear speed in test S3. This is shear zone. In this way, coarse grains tend to concen-
probably due to the fact that test S2 had greater excess pore trate in the vicinity of the shear zone. Consequently, small
pressure during shearing, which led to lower effective particles tend to remain above and below the shear zone,
normal stress, thus less progress in grain crushing. and at the same time small particles tend to drop into
A general tendency is that greater amount of SS-par- the pores between the larger grains due to gravity. In
ticles is observed in the lower shear box in the three tests. addition to this process, some of the small particles dropped
Although deposition of SS-particles was observed when through voids at the moment shearing was terminated.
shearing was stopped, the ratio was still greater than 1.0 Distribution of small particles after the test is the result
at heights of about 3 cm and 4 cm above the shear zone in of these effects.
test S2. This suggests that the finest particles generated Wafid et al. 2004 studied the segregation process during
by grain crushing were not only deposited downward, ring shear tests. In their study, undrained ring shear tests
but also moved upward. Sassa et al. (2004a) observed that were conducted on Silica sand No. 8 (D50 = 0.048 mm) un-
silty water generated by grain crushing during ring shear der normal stress of 180 kPa by using the shear-stress-
test under high speed high normal stress was diffused controlled mode. The maximum shear speed was limited
both upward and downward from the shear zone. The to 5.4 cm s1 and the samples were sheared up to a dis-
smallest particles floating in the pore water have also placement of 1000 cm. They found that large grains con-
diffused. This process may contribute to the diffusion of centrated in the upper part of the shear zone and small
the finest particles. particles accumulated at the bottom of the shear zone
As for the variation of LL-grains distribution repre- when the sample reached the steady state. They called this
sented in red lines, the ratio shows greater than 1.0 for phenomenon segregation. When precisely observing the
tests S1 and S2 in the vicinity of the shear plane (more distribution of large grains in Fig. 7.15, the large grains
precisely described as at the level of the shear plane and seem to be accumulated in the upper part of the shear
around 1 cm above the shear plane in Fig. 7.15). It means zone. The result appears consistent with that presented
that LL-grains concentrated in the vicinity of the shear by Wafid et al. 2004.
110 Hiroshi Fukuoka Kyoji Sassa Gonghui Wang

7.7 Conclusions Acknowledgments

By using a newly developed ring shear apparatus with a Mr. Naohide Ishikawa, and Mr. Ryo Sasaki former master
transparent shear box, a series of naturally drained mono- course graduate students are acknowledged for conduct-
tonic shear-speed-controlled tests were conducted on ing the preliminary study for this research in our labora-
mixed sand and mixed beads to study the process of shear tory. The authors deeply appreciate the members and stu-
zone development. Movement of grains and silty water dents of the Research Centre on Landslides, Disaster Pre-
due to grain crushing in the samples was monitored by vention Research Institute, Kyoto University for their as-
video cameras. The variation of pore water pressure and sistance to this research.
shear resistance were monitored. The variation of grain This study is a part of the M101 Project, Areal Predic-
size distribution in the sample was examined for sliced tion of Earthquake and Rain Induced Rapid and Long-
samples after tests for quantitative analysis of vertical traveling Flow Phenomena (APERITIF, proposer:
displacement of grains and particles as well as creation K. Sassa), of the International Programme on Landslides
of fine particles due to grain crushing. The following re- (IPL) operated by the International Consortium on Land-
sults were obtained: slides (ICL) since 2002, and is also a part of the Japanese
project Aerial Prediction of Earthquake and Rain-In-
1. In the naturally drained ring shear tests under greater duced Flow Phenomena (APERIF project, represented by
shear speed and greater normal stress conditions, K. Sassa), which was supported by the Special Coordinat-
greater excess pore pressure and lower apparent fric- ing Fund for Promoting Science and Technology of the
tion were observed. Comparing the test results with dif- Ministry of Education, Culture, Sports, Science and Tech-
ferent conditions of normal stress (200 kPa and 22 kPa) nology (MEXT) of the Japanese Government for 3 years
and different shear speeds (150 cm s1 and 9 cm s1), the from 20012004.
greatest pore pressure build-up and the minimum ap-
parent friction were obtained in the combination of
high normal stress and high speed shearing test. It is References
probably because high normal stress contributes to
great grain crushing and the resulting volume reduc- Bridgwater J (1972) Stressvelocity relationships for particulate sol-
tion, and high speed contributes to a high pore pres- ids. ASME Paper No. 72-MH-21, 7 p
Fukuoka H (1991) Variation of the friction angle of granular mate-
sure generation rate and the resulting greater pore pres-
rials in the high-speed high-stress ring-shear apparatus, influ-
sure build-up. ence of re-orientation, alignment and crushing of grains during
2. Grain size analysis of the samples sliced layers after shear. Annu Disaster Prev Res Inst Kyoto Univ 41(4):243279
the tests revealed that greater amounts of small par- Fukuoka H, Sassa K, Shima M (1990) Shear characteristics of sandy
ticles were generated under greater normal stress. soils and clayey soils subjected to the high-speed and high-stress
ring shear tests. Annu Disaster Prev Res Inst Kyoto Univ 33(1):
Along with the diffusion of silty water generated by
179190 (in Japanese)
grain crushing, small particles were transported up- Fukuoka H, Sassa K, Wang GH, Sassaki R (2005a) Measurement of
ward and downward from the shear zone. velocity distribution profile in ring-shear apparatus with a trans-
3. The concentration of large grains from the center to parent shear box. In: Sassa K, Fukuoka H, Wang GH, Wang F (eds)
the upper part of the shear zone was confirmed by Landslides, risk analysis and sustainable disaster management.
means of visual observation through the transparent Springer-Verlag, Berlin Heidelberg New York, pp 149156
Fukuoka H, Sassa K, Wang GH, Sassaki R (2005b) Observation of
shear box together with grain size analysis of the sliced shear zone development in ring-shear apparatus with a trans-
samples from several layers after the test. On the other parent shear box. Landslides (in contribution)
hand, small particles were accumulated mostly below Hungr O, Morgenstern NR (1984) High velocity ring shear tests on
the layer where large grains were accumulated. The sand. Gotechnique 34(3):415421
reason why large grains were accumulated near and in Ishikawa N (2004) An experimental study on the shear behavior of
granular materials by means of a new ring shear apparatus with
the shear zone could be interpreted as follows: grains
a transparent shear box and image processing. Masters thesis,
under shearing are subjected to vertical movement, the Kyoto University
penetration resistance of larger grains into a layer of Lang Y, Ote K, Fukuoka H, Sassa K (1991) Image-processing the ve-
moving particles is smaller than that into the static locity distribution of grains in ring shear tests. In: Proceedings
layer. Therefore, both lager grains below the shear zone of the Conference of Japan Society of Erosion Control Engineer-
ing 1991, pp 302305 (in Japanese)
and larger grains above the shear zone tend to move
Lemos LJL (2003) Shear behaviour of pre-existing shear zone under
into the layer of moving grains in the shear zone. At fast loading insights on the landslide motion. In: Proceedings
the same time, smaller particles can drop into the pores International Workshop on occurrence and mechanisms of flow-
between the larger grains downward due to gravity. like landslides in natural slopes and earthfills, pp 229236
Chapter 7 Shear Behavior and Shear Zone Structure of Granular Materials in Naturally Drained Ring Shear Tests 111

Melosh HJ (1986) The physics of very large landslides. Acta Mech Sassa K, Fukuoka H, Scarascia-Mugnozza G, Evans SG (1996) Earth-
64:8999 quake-induced-landslides: distribution, motion and mechanisms.
Novosad J (1964) Studies on granular materials. II. Apparatus for Special issue for the great Hanshin Earthquake Disasters, Soils
measuring the dynamic angle of internal and external friction and Foundations 5364
of granular materials, Collection Czech Chem Commun 29: Sassa K, Wang G, Fukoka H (2003) Performing undrained shear tests
26972701 on saturated sands in a new intelligent type of ring shear appa-
Sasaki R (2005) An experimental study on the process of shear-zone ratus. Geotech Test J ASTM 26(3):257265
development using the ring shear apparatus with a transparent Sassa K, Fukuoka H, Wang GH, Ishikawa N (2004a) Undrained dy-
shear box. Master Thesis of Graduate School of Science, Kyoto namic-loading ring-shear apparatus and its application to land-
University, 74 p slide dynamics. Landslides 1:719
Sassa K (1984) The mechanism starting liquefied landslides and Sassa K, Wang G, Fukuoka H, Wang F, Ochiai T, Sugiyama M, Sekiguchi
debris flows. In: Proceedings of 4th International Symposium on T (2004b) Landslide risk evaluation and hazard zoning for rapid
Landslides, Toronto, June, 2, pp 349354 and long-travel landslides in urban development areas. Landslides
Sassa K (1988) Geotechnical model for the motion of landslides. In: 1(3):221235
Special Lecture of 5th International Symposium on Landslides, Scarlett B, Todd AC (1969) The critical porosity of free-flowing sol-
Landslides, 1015 July, 1, pp 3755 ids. J Eng Ind Ser A 91(1):478488
Sassa K (1996) Prediction of earthquake induced landslides. In: Pro- Tika TM (1989) The effect of rate of shear on the residual strength of
ceedings of 7th International Symposium on Landslides, A.A. soil. PhD thesis, University of London (Imperial College of Sci-
Balkema, Trondheim, 1721 June, 1, 115132 ence and Technology), 494 p
Sassa K (2000) Mechanism of flows in granular soils. In: Proceed- Tika TE, Hutchinson JN (1999) Ring shear tests on soil from the Vaiont
ings of the International Conference of Geotechnical and Geo- landslide slip surface. Gotechnique 49(1):5974
logical Engineering, GEOENG2000, Melbourne, 1, pp 16711702 Vibert C, Sassa K, Fukuoka H (1989) Friction characteristics of granular
Sassa K, Fukuoka H, Lee JH, Zhang DX (1992) Measurement of the soils subjected to high speed shearing, In: Proceedings the Japan
apparent friction angle during rapid loading by the high-speed China Symposium on Landslides and Debris Flows, 1, pp 295299
high-stress ring shear apparatus; interpretation of the relation- Wafid MA, Sassa K, Fukuoka H, Wang GH (2004) Evolution of shear-zone
ship between landslide volume and the apparent friction during structure in undrained ring-shear tests. Landslides 1(2):101112
motion. In: Proceedings 6th International Symposium on Land- Wang GH, Sassa K (2002) Post mobility of saturated sands in undrained
slides, pp 545552 load-controlled ring shear tests. Can Geotech J 39(4):821837
Chapter 8

Rockslides and Their Motion

Mauri McSaveney* Tim Davies

Abstract. The motion of landslides sourced from mostly bedrock (called flow, but which fell on and mobilized a fine, saturated sub-
rockslides) is controlled by the phenomenon of grain flow, and the strate. Field and physical and numerical modeling stud-
frictional resistance of the constituent rock grains and their intersti-
ies have been used to gain understanding of the motion
tial fluids. Modern understanding of grain-flow dynamics recognises
that the important interactions between grains are irregularly dis- of these rockslides during emplacement.
tributed within the grain mass, with fortuitous alignments of grains
carrying most of the stress in force chains,while other grains are only
weakly stressed. In rapidly shearing grain flows, under substantial 8.2 What is a Rockslide?
confining stress, force-chain stresses rise high enough to crush grains.
Such comminuting grain flows develop a distinctive grain-size distri-
bution that is fractal over many orders of magnitude of grain size David Varnes (Varnes 1978, updated in Cruden and Varnes
down to sub-micron sizes. In the moment of crushing, grains are not 1996) presented the most widely adopted landslide clas-
solids, and behave as high-pressure fluids. As the grain fragments are sification. It is useful because it is simple and adaptable,
injected into lower pressure surroundings, they behave as would any and not because it is completely accurate; no classifica-
other fluid, lowering the effective stress on other grains, and thereby tion can be perfect, but a common classification is essen-
lowering frictional resistance to flow. We show how this affected the
blockslide component of New Zealands prehistoric giant Waikaremo-
tial if people are to know that they are discussing similar
ana rockslide; New Zealands Falling Mountain rock avalanche trig- things. The terms rock slide and rockslide are synonyms.
gered by an earthquake in March 1929; and a small prehistoric New They imply a mass, or contiguous group of masses of
Zealand rockslide that was too small to be a comminuting grain flow, mostly bedrock that moved or is moving downslope, ap-
but which fell on and mobilized a fine, saturated substrate. We use parently dominantly on surfaces of rupture, or on rela-
grain-flow dynamics to explain the motion of these rockslides deter-
tively thin zones of intense shear strain, so that the dis-
mined through field studies and physical and numerical modeling.
placed mass of bedrock appears to have slid. The displaced
Keywords. Rockslide, rock avalanche, grain flow, comminution, fragmen- mass may be somewhat disrupted, and there may also be
tation, fracture energy, force chain, grain bridge, frictional resistance evidence of some distributed deformation (flow) within
it (Fig. 8.1). Rockslides become debris slides when the
displaced mass appears completely disrupted, and this
8.1 Outline transition can occur in time or in space on the same land-
slide, so that what might technically be debris slide in the
The larger landslides on the rocky planets invariably in- strict Varnes classification may still be called a rockslide
volve bed rock, and are generally called rockslides, which in practice. There also can be transitions from sliding to
are the largest natural manifestations of the phenomenon flow or spreading (translation of blocks on a mass with
of grain flow. Here, we first describe what is meant by the distributed flow). Since the transitions can be gradational,
term rockslide, and discuss modern understanding of and not necessarily readily recognisable, what may be a
grain-flow dynamics as a process governing the motion rockslide to some, may be a debris flow or a rock spread
of landslides. Of particular interest in large rockslides is to others depending on the details of their perception of
the distinctive grain-size distribution that results from what is important. This seldom presents any particular
comminuting grain flow. We discuss this grain-size dis- technical difficulty in understanding as it has no bearing
tribution, how it arises, and its significance in the ener- on the mechanical processes of the motion.
getics of rockslide motion. Last we look at how commi- At the larger end of the rockslide scale, are what are
nuting-grain-flow dynamics has affected some real often termed rock avalanches or rockslide avalanches in
rockslides: the blockslide component of New Zealands the literature (Fig. 8.2). Avalanche is not in the Varnes clas-
prehistoric giant Waikaremoana rockslide; New Zealands sification, but is acceptable in English usage. The term
Falling Mountain rock avalanche triggered by an earth- avalanche has never necessarily implied snow or ice, but
quake in March 1929; and a small prehistoric New Zealand in the context of snow avalanches, the qualifying adjec-
rockslide that was too small to be a comminuting grain tive snow is normally readily apparent in context and is
114 Mauri McSaveney Tim Davies

Fig. 8.1.
Aerial view of the Waikaremo-
ana landslide and a relief map.
a Aerial view of the Waikare-
moana landslide showing where
the blockslide component plowed
into the moving rock avalanche
and pushed up a 150 m high
mound of debris as it came to
rest from a speed of 40 m s1
(photograph by Lloyd Homer);
b relief map of New Zealand
showing locations of the main
rockslides described in the text

dropped. If the greater part of the displaced mass trans- the rock mass, but more usually the sliding surface is nei-
lates as an apparently intact block, it may be called a block ther circular nor planar, and some amount of rock-mass
slide or blockslide (Fig. 8.1), or even a block glide. These break-up is inevitable due to flexing of the mass.
are just combinations of simple English words, each with Although frictional sliding often appears to be the
precise meaning given by context. Rockslide avalanches, mechanism of displacement of rockslides, this is never
and blockslides are all encompassed by rockslide. true. The displacement mechanism is always grain flow,
The apparent sliding movement can be rotational and the deformation takes place across a zone of granu-
(moving on a concave curved surface) or translational lar material of finite thickness. For this reason, as with all
(moving on a planar surface). If rotation is along a circu- landslides, a key to understanding rockslide motion is
lar arc, it can be accomplished with minimal break-up of understanding grain-flow mechanics.
Chapter 8 Rockslides and Their Motion 115

Fig. 8.2.
The Falling Mountain rock ava-
lanche triggered by an earth-
quake in 1929 (photograph by
Lloyd Homer). The landslide is
believed to have first begun to
move as a blockslide, which rap-
idly accelerated as it disinte-
grated to joint-bounded blocks
on flexing during evacuation of
the scar, before comminuting to
mostly sand and silt in its 4 km

exist in all grain flows where grains are moving towards

8.2.1 Force Chains and Their Role in Grain-flow Mechanics
one another; they cannot exist where grains are diverging,
because this brings grains out of contact and breaks the
The deforming part of all landslides is always made up of force chain. Force chains can not normally be seen. Experi-
granular material, with an interstitial fluid of some kind, mental two-dimensional grain flows using uniform discs
most often air or water. It is through deformation of this that polarize transmitted light under stress have been used
granular material that all landslides move, and it is through to image force chains (Fig. 8.3) (Howell, et al. 1999a,b), and
the interaction of the grains and their interstitial fluids that they have been modeled numerically in four dimensions
granular materials gain their frictional properties exhib- for materials of uniform and widely graded grain sizes
ited during flow. Granular friction resists motion, so fric- (Hazzard and Mair 2003; Mair et al. 2001).
tional properties determine whether a given landslide sub- Present in any grain mass, whether moving or static,
ject to a given force speeds up or slows down. Grain flows force chains form a 3-dimensional mesh of connected
are of great industrial importance; they are much re- more highly stressed grains forming relatively short
searched, but not generally well understood (Campbell beams, columns and braces (to use a building analogy).
2006). The literature is diverse and spread across many As the range of grain sizes present in a grain flow increases,
disciplines, in many languages, and with diverse techni- force chains become more diffuse, tending to proliferate
cal jargon to inhibit cross-discipline communication. around larger grains, but they still exist. All experiments
Long-recognised complexities of industrial grain flows, show that much of the resisting stress in a shearing grain
are that flow can lock up unpredictably, and be difficult to flow is carried by force chains. The principle compres-
restart, or the stopping of flow can transfer the flows mo- sive stress within each grain is highly nonuniformly dis-
mentum to the container and rupture it. These occur be- tributed among grains, and strongly controlled by the
cause, although the grain mass as a whole behaves as a fluid, shear rate. Some grains carry little of the average prin-
it is made up of rigid grains, and when deformation ran- ciple compressive stress, while others carry much of it, so
domly assembles a series of grains into perfect alignment some grains are highly stressed. For shear strain to per-
in direct contact, the chain of grains is a solid rod that re- sist, the force chains must fail, and force chains are con-
sists deformation until it fails for some reason. These chains stantly forming, stressing, failing and reforming among
are called force chains, or grain bridges (Campbell 2006; other grains in all grain flows. Where the flowing grains
Liu et al. 1995; Sammis et al. 1987). A well known example are more heterogeneous in size, shape, or composition,
of a usually static force chain is the stone arch. Force chains the stress field of force chains is more varied.
116 Mauri McSaveney Tim Davies

kinetic energy of its fragments moving away from its

former center of mass.
Several elements of force-chain failure are of particu-
lar interest in landslide behavior. If the granular mass has
significant cohesion through the presence of clays and
water, force chains can exist in tension, and so the diver-
gence of grains (as well as their convergence) is resisted.
Hence such cohesive materials can provide more resis-
tance to flow than were they dry and non-cohesive. In
addition, the platy character of clay minerals makes for
particularly weak force chains as grains become aligned
by shear displacement, and slip is facilitated by water films
and grain alignments.
Force-chain induced fracture of brittle grains is im-
portant in rockslides. The strength of materials increases
as the rate of application of stress increases; the faster a
landslide mass deforms, the more stress its grains can bear
before failure, and the transfer of stress at grain failure
becomes higher. In part, this is because as the strain rate
Fig. 8.3. Transparent plastic discs shearing under stress viewed with increases, the amount of super stress above the grain yield
transmitted circularly polarised light to show force-chain stress con- strength increases because of the finite time for fracture
centrations (http://www.phy.duke.edu/~bob/). This false-color image crack initiation and growth. This increases the probabil-
is from Dan Howell (Howell et al. 1999a,b). This is a 2D experiment
ity of rapid, catastrophic failure, and decreases the mean
in which a collection of disks undergoes steady shear. When viewed
through circular polarizers, high forces show up as bright regions. fragment size when fracture occurs. When individual
When the force is very large, the polarization is rotated through grains crush, they momentarily cease being solids; they
multiple phases of , showing fringes. The experiment shows that can bear no directed stress, and briefly behave as a fluid
forces in granular systems are inhomogeneous and intermittent if under pressure, transmitting an isotropic stress (pressure)
the system is deformed. The forces are detected by means of
to their surroundings. The value of this pressure is the
photoelasticity: when the grains deform, they rotate the polariza-
tion of light passing through them. In granular materials, force is applied directed stress causing grain fracture at the in-
rarely transmitted uniformly, but rather preferentially along a net- stant of grain failure, it is always a little higher than the
work forming force chains often right next to regions where there compressive strength of the grain material under the
is little or no force. The illustrated disks are under a large load the ambient conditions of confining stress, temperature, and
high-force areas are highlighted in red (false color). Note that some strain rate, but it diminishes as the crushing grain expands
of the disks are not coloured at all (no force), while disks right next
to them are highlighted in red (high force)
into its lower-pressure, surrounding grain mass beyond
the immediate failing force chain. Fragments from the
crushing grain are pushed out of the shortening force
Individual force chains are usually short. There is no chain, and injected into surrounding lowerpressure re-
theoretical reason for any particular length, although gions, between other, less stressed grains, thereby forcing
lengths of one and two grains are meaningless and pro- them apart, and transferring stress from the force chain
hibited by semantics. Although the meshwork must to surrounding grains.
straddle the grain mass, the alignments are fortuitous, When ephemeral force chains fail, their temporarily
and so long perfect alignments are rare, and very short stored elastic energy returns to kinetic energy, but as in
chains are indistinguishable from nodes in the mesh. In all such things, some energy is lost to heat. With slip, ro-
modeled grain flows, chain lengths are variable, but gen- tation and buckling, kinetic energy is returned only in
erally ~515 grains. the direction of slip, rotation or buckle. In grain break-
Force chains can fail by several mechanisms; most fail age, it is returned in all lower-pressure directions radial
by inter-granular slip, grain rotation, or buckling of the to the fragmenting grain. That is, momentarily, the crush-
whole chain, some can fail by grain divergence, but the ing grain transfers a stress equal to its compressive
weaker and more highly stressed grains can fail by crush- strength under the ambient loading conditions, from the
ing (grain fracture) when force-chain stress exceeds grain direction of the principle compressive strain to all other
strength. At failure, the elastic strain stored in all the stress components, resulting in overall increases in the
grains participating in a force chain is released, at least in mean regional stresses in these directions in proportion
part, and this drop in elastic potential energy appears as to the volume of grains fragmenting, and the compres-
kinetic energy in the direction of failure as rotating and sive strength of the rock (which varies with confining
(or) translating grains, or if a grain has fractured, as the stress and rate of application of load). By adding to the
Chapter 8 Rockslides and Their Motion 117

mean stress that supports the overburden weight, the ef- (8.3)
fective overburden weight carried by other grains is re-
duced, and they can move with reduced frictional resis- were a and b are empirical constants.
tance. Lateral and longitudinal mean stresses are also in- Distribution by mass, however, is deceptive and is an
creased, and these also can significantly affect motion in inappropriate measure when grain mass is not of inter-
thick crushing grain flows. est. Of interest in grain-to-grain interactions are the size
The effect of grain crushing is simplest understood of the average grain and the probabilities of grains en-
through the well known Mohr-Coulomb relationship countering grains of similar or different sizes. Probabil-
(Eq. 8.1) which applies also to non-fragmenting granular ity of grain encounters is linearly proportional to grain
flows. It relates shear stress, , to grain contact stress surface area, not grain volume: the more total grain sur-
through an internal friction angle (Eq. 8.2) or coeffi- face area is contained on grains of a given size, the more
cient of internal friction such that: likely that a given grain of any size will make contact with
a grain of that size. The total surface area on grains of a
= ( p) S0 (8.1) given size is determined by the average surface area of
such grains, and their total number. If we assume that all
and grains have a similar range of shapes, regardless of size,
then grain surface area (ad) of a given grain size (dia-
= tan (8.2) meter d) will be given by:

for intergranular contact (effective) stress ( p) ( is ad = s d2 (8.4)

overburden load and p is pore-fluid pressure) and cohe-
sive strength S0. The crushing grains add an ephemeral where s is a shape factor >1.0 relating the surface area of
high fluid pressure p to the grain mass and reduce the the average grain shape to the area of a sphere of the same
effective confining stress on grains in proportion to the mass. As long as our interest is only in relative areas, and
crushing strength of the material and the proportion of all grain sizes have similar ranges of shapes, we do not
the grain mass that is crushing simultaneously. Since the need to know this shape factor. The total area (Ad) con-
crushing strength is usually very much higher than the tained on all grains of a given size per unit volume of
overburden stress (for typical brittle crustal rocks, the deposit is given by:
pressure of a fracturing grain can be of the order of 0.5 to
5 GPa, and much more than the overburden stress in even Ad = nd ad (8.5)
the largest landslides) only a small proportion of simul-
taneous crushing grain mass is needed to significantly where nd is the number of grains in that volume of a given
lower the frictional resistance to flow. size d. The total grain surface area A then is:
Crack growth and hence grain fracture are catalysed
by water, and therefore occur at slightly lower stresses than
in dry rock, a difference noticeable mostly on the water-
less planetary bodies of Mercury and the Moon. Of course,
under water-saturated conditions, the pore-water pressure Mean grain size (dm) per unit volume is given simply
is additional to that from grain fragmentation, and a lower by:
proportion of simultaneously crushing grains will pro-
duce the same reduction in frictional resistance.

8.2.2 Grain-size Distribution in Rockslides where is the ratio of densities of deposit and grain ma-
terial (ignoring for simplicity, density differences between
In the conventional measure of grain-size distribution component mineral grains which become more promi-
(distribution by mass) the grain-size distribution in many nent at the smallest sizes).
landslides appears to follow a two-parameter Weibull dis- Critical to accurate estimation of mean grain size, and
tribution (Eq. 8.3), the so called Rosins Law of crush- relative proportions of surface area is accurate knowledge
ing, or Rosin-Rammler distribution) well enough to pass of the complete grain-size distribution, particularly in the
some statistical fitting tests, but its usually slight misfit smallest grain sizes where most of the grains are. A grind-
appears to be systematic across the range of grain sizes ing limit of about 1 m for quartz has been suggested
(McSaveney 2002; Dunning 2004). In the Weibull distri- (e.g., An and Sammis,1994). Electron-microscopy on a
bution, the proportion by weight of rock debris (Pf) finer number of rock-avalanche samples (A.L. Strom, RAS
than a given grain size x, is given by Moscow, personal communication 2005), includes several
118 Mauri McSaveney Tim Davies

recent New Zealand examples supplied by us (1991 Mount major differences in sample histories, both samples are
Cook, and 1992 Mount Fletcher II rock avalanches identically self-similar across the range of grain sizes
(McSaveney 2002)), and provides convincing evidence of above 0.6 m. (and almost identically non-fractal below
grains in sub-micron sizes down to the resolution of the 0.6 m). Electron-microscope images of debris-avalanche
electron microscope used for counting (~0.077 m). grains (such as Fig. 8.9c,d of Belousov et al. 1999) reveal
Grain counts across two size ranges (~0.0772.75 m a possible explanation for the change in distribution trend
and ~1.010 m) have unknown quantitative relation- below 0.6 m. Such images always show sub-micron frag-
ships to each other and to grain counts calculated from ments clinging to sand grains. Increasingly smaller grains
conventional weight-percentage particle-size analysis are increasingly more influenced by Van der Waals forces,
over the range 0.002100 mm. By assuming that the sub- and adhere strongly to any other grain; hence they are
samples are parts of the same grain-size distribution, removed on larger grains if these are selectively removed
however, the three distributions were aligned by simple in analysis. Below an undetermined size smaller than
multipliers to approximate the complete distribution 0.077 m, smaller sizes are incapable of maintaining in-
(Fig. 8.4). Both the Mount Cook and Mount Fletcher II dependent existence; they adhere too strongly to other
samples are sufficiently similar over overlapping portions grains, and so are not measurable after the event. If these
of the range of sizes, and so no error is likely to have arisen are the causes of the reversal in trend at the smallest sizes
by assuming that the distributions should be continuous in these grain-size distributions, it implies that a grind-
from one sub-sample to the next. ing limit recognised in industrial grinding is not the
Above about 0.6 microns, both distributions closely smallest size of fragments capable of being created, but
follow a simple Power-law distribution with an exponent the smallest size capable of maintaining independent ex-
of ~2.58, to the largest grains present in the deposits with istence. Much smaller sizes are possible, and may be pro-
no systematic departures. That is, the distributions are duced, but they can not be verified by this technique.
fractal across most of the range of sizes present. The Hereafter, we assume that grain comminution is identi-
fractal nature of comminution by fragmentation is well cally self-similar across the entire range of particle sizes
established (Turcotte 1986). The fractal character has sig- present in grain flows, down to and including the small-
nificant implications as to the origins of deposits and the est sizes capable of maintaining independent existence.
dynamics of the processes that created them. This suggests that at 0.077-m nominal grain diameter
The sub-micron portions of both distributions are non- 99.8299.98% of the grains of this size might have been
fractal (Fig. 8.4). There are systematic decreases in pro- missed in counting, for one reason or another, but if so,
portions of grains below 0.6 m in both deposits. Despite the missed grains account for less than 0.1% of the mass
(<1 g kg1). The assumption is one of convenience; it is
not essential to understanding the role of comminution.
Grain flows in which grain crushing has been impor-
tant are thus very distinctive. Not only are they much finer
grained than their parent materials, they also exhibit a
distinctive grain-size distribution that makes them look
as if their grains have never been sorted by mass or volume.
Since all sizes are present, they can never have been selected
for size in their making, other than to have become as small
as possible in the time available, but the apparent lack of
sorting is deceptive, because it demonstrates that a sort-
ing mechanism has operated (see below).
Although crushing-grain-flow deposits often appear
to be what is termed matrix-supported (the classical
Blocks-in-Matrix or BIM rocks), apparently matrix-sup-
ported grain masses can grade laterally and vertically to
apparently clast-supported debris, usually with no appar-
Fig. 8.4. Distribution of grain numbers by grain size in two recent ent reason, and sometimes with identifiable inherited
New Zealand rock avalanches (Aoraki/Mount Cook of December source structures traceable across the gradational bound-
1991 and Mount Fletcher II of September 1992, McSaveney 2002). aries. On detailed sampling and size analysis, relative-size
Grain counts courtesy of Dr. A.L. Strom, RAS, Moscow. The line is relationships between larger and smaller grains are self-
drawn with a power-law exponent of 2.58; it has not been fit to the similar at most scales, and geometrically the deposits ei-
data. We explain the departure of the data from the power-law at the
finest sizes as resulting from the finest grains selectively adhering to
ther are all matrix, bar the largest few grains, or have no
other grains through Van der Waals forces and being selectively de- matrix. The difference between matrix- and clast-sup-
pleted during grain-size analysis ported structures in comminuted grain masses is merely
Chapter 8 Rockslides and Their Motion 119

one of perception and degree of comminution, which of-

ten is spatially very highly variable in these deposits. Lo-
cally lesser comminution can arise because: the locally
viewed grain mass is from a larger grain crushed late in
the flow; the rate of last grain crushing was slow; the local
shear rate was always slow; or any combination of these
only partly independent variables. The number of frag-
ments from a single crushing is a positive power-law func-
tion of strain rate (Zhang et al. 2004).
In theoretical and experimental studies, Sammis et al.
(1987) found fault-gouge grain-size distributions to be
characterized by a fractal dimension of 2.58 (based on a
theoretical three-dimensional geometry with no two
grains of the same size adjacent to one another, although
this arrangement is impossible to demonstrate by direct
observation). Other theoretical models with different
fractal dimensions exist (1.97, Turcotte,1986; 2.84, Allegre
et al. 1982). In addition to the two samples discussed
above, we (Davies and McSaveney, in press) find that
fractal dimensions of a number of grain-size distributions
from pervasively crushed samples from many rockslide
avalanches and two fault gouges (Fig. 8.5) are well approxi-
mated by 2.58. In more detailed sampling of a few rock-
avalanche deposits, some in common with the data of
Fig. 8.5, Dunning (2004) finds a variety of lower fractal
dimensions that appear to approach and reach 2.58 with
sample distance from source. Crosta et al. (in press) find
a variety of fractal dimensions ranging from 1.2 to 2.9 in
the Italian Val Pola rock-avalanche deposit. Possibly of
most significance in these studies is the conclusion that
grain-size distributions by number are almost always
fractal across most of the range of grain sizes present in
such deposits they do not follow Rosins Law, but they
are formed by crushing.
The Val Pola evidence may imply that several mecha-
nism are causing fragmentation, and acting to greater or
lesser extents depending on local conditions, such as grain Fig. 8.5. Grain size distributions. a Grain-size distributions of a num-
confinement, and grain material. Possible fracturing ber of rockslide deposits expressed conventionally as proportion by
weight finer than a given grain size. Alpine Fault and Casey are
mechanisms are crushing between grains, grain-to-grain samples of fault gouge. b Grain-size distributions of the same rockslide
collisions (impacts), attrition of grain edges, and grain deposits expressed numbers of grains per unit volume of a given grain
bending moments. All involve locally exceeding the size. The power-law line is drawn with an exponent of 2.58
strength of the material. At low confining stress, large
voids additionally allow gravitational mechanical sieving size, and all are therefore likely to be of different sizes. Grain-
to change grain-size distributions, but in general, fractal size crushing appears to be such a mechanism, but it can not
distributions and high confining stresses strongly inhibit directly produce more finer grains than coarser grains in a
gravitational sieving. Immense dust clouds are usually seen single fracturing event, this can only evolve through many
with rockslides, and it is likely that this is expelled from the events and with other processes operating.
weakly confined, usually coarser outer part of the rockslide, Comminution is progressive, and there is a fixed lower
that we have come to call the carapace (Dunning 2004). limit below which grains do not maintain independent
The extremely dangerous environment during rock ava- existence, so the total number of grains in any given vol-
lanching precludes direct field observation of any mecha- ume of debris must increase with continuing comminu-
nism, prohibiting evaluation of relative roles. tion by increasing the numbers of smaller grains at the
The simplest theoretical model which has no two grains expense of the largest grains (this is axiomatic from the
of the same size adjacent to one another is one in which no definition of grain comminution). As the number of
two grains anywhere have any reason to be exactly the same grains increases, the proportion represented by the larg-
120 Mauri McSaveney Tim Davies

est grain must decline, even as the largest grain size de- interactions is proportional to grain-surface area (Fig. 8.6),
clines (since there is always only one largest grain). Hence and most grain interactions most of the time are between
one might expect the power-law exponent of the size dis- the grain sizes that represent most of the grain-surface
tribution by number of grains to decrease (become more area, and not with the grains that represent most of the
negative) with continuing comminution once the modal volume. Since grain-flow deformation takes place by rela-
size approaches the grinding limit. Hence, there should tive motion between grains, it follows that during much of
be no preferred fractal dimension that such size distribu- the flow duration, the deformation in these comminuting
tions could evolve to. If there is a preferred fractal dimen- grain flows takes place between smaller grains at all scales.
sion around 2.58, it has to arise in tandem with the grain- Because the comminution process appears to be iden-
creation mechanism. That is, if there are no two adjacent tical at all scales, we do not necessarily need to study the
grains of the same size, it is because the grain-creation pro- process at micron and sub-micron scales; it suffices to
cess is simultaneously destroying such fortuitous combi- comprehend it at any macro-scale, and to verify that scale
nations because there is no process especially creating such is not a relevant parameter at macro-scales. It also fol-
combinations (ie there is no preferred size, and so all sizes lows from the fractal distribution that every grain, even
are made). We know also that there is no process creating in what appears to be merely the dust among the larger
larger grains, except at the grinding limit which we do not grains, is as important to the flow dynamics as is any other
see. Of course, all grain flow processes continually rearrange grain in the flow, and the dust should not be ignored sim-
grains so that similar-sized grains can always fortuitously ply because it is unseen.
come into contact. They will do this more frequently than The least relevant grains are the large boulders that
they can arrange themselves into 515 grain force chains, often litter the deposit surface. They are less relevant be-
so it must occur frequently. Sammis et al. (1987) consid- cause they lack the same opportunities to interact with as
ered that combinations of similar-sized grains were de- many surrounding grains as do grains in the interior. They
stroyed preferentially, but there is no valid theoretical rea- participate mostly by providing overburden pressure. A
son for this to occur, rather, smaller grains should be pref- carapace of larger grains is commonly found on the up-
erentially crushed between larger grains because less de- per surface of many rockslide deposits. Carapaces can
formation, and hence less work is required to crush them. vary in thickness, but generally are about 10 m thick. Indi-
In the two samples for which we have full grain size vidual carapace boulders can range up to megaclasts hun-
distributions, 99.5% of all grains are <10 m, so not sur- dreds of meters across. Large Toreva blocks of original
prisingly, the mean grain sizes of these finely comminuted volcanic edifice sometimes kilometers across are often rec-
deposits are similar and ~1 m. By grain number, they ognised on top of volcanic debris avalanches, and some large
are mostly fine clay, whereas by grain weight, they would blockslides cover tens of km2 with mostly intact blocks hun-
be described as bouldery, silty sand. Since boulders to dreds of meters thick, and only a relatively thin layer (a few
more than 10 m across are present in the deposits, clearly meters thickness) of comminuted grains at their base.
smaller grains have been crushed preferentially.
Although we lack electron microscopy for our other
samples, they are all very similar across the range of sizes
determined (down to 2 m for several samples), strongly
suggesting that they too might extend similarly to below
1 m (but their distributions below 0.6 m are as yet un-
known). Few if any deposits have no visible dust, and ev-
ery dry rockfall is accompanied by a prominent dust cloud
that limits detailed observation. Calculation of accurate
mean grain size and accurate proportions of surface area,
are critically dependent on knowing the distribution in
the range below 1 m. In any event, the deposit mean grain
sizes are likely to be very small, and most (>99%) of the
grains in these comminuted-grain-flow deposits are silt and
clay-sized particles. This is well hidden from view by the
<1% of larger grains that occupy almost the entire vol-
ume, and lead such deposits to be described as bouldery,
at least until their interiors are exposed by erosion. Fig. 8.6. Distribution of total grain surface area by grain size for the
Any model of how this peculiarly regular grain-size rockslide deposits of Fig. 8.4. The power-law line is fit to the Mount
Cook data of Fig. 8.5 which was determined from a much larger
distribution comes to exist must entail interactions be- sample than those available for Fig. 8.4, and so is considered to be
tween grains, either by considering real grain interactions more representative of the larger fraction of the Aoraki/Mount Cook
or considering them by proxy. The probability of grain rockslide avalanche
Chapter 8 Rockslides and Their Motion 121

formation is that the distribution is self-similar at all

8.3 Size Distribution from Grain Crushing
scales, and presumably formed by the same processes at
all scales. Also, injection of finer grains between larger
It is self-evident that mass is conserved in grain crushing, grains confirms that the fragmentation process can be
so the sum of fragment volumes must remain the origi- viewed as introducing a high-pressure fluid into the pore
nal volume of the crushed grain. It is observed empiri- space; this fluid competes for pore space with every other
cally that the number of fragments is a positive power- fluid that might be present.
law function of strain rate (Zhang et al. 2005). Fragments
in crushing are defined by the intersections of cracks as
they propagate through the crushing grain. There appear 8.3.1 Energy Loss in Rock Comminution
to be no other contraints. The largest fragment must be
smaller than the parent; the smallest fragments may be Such intense rock comminution as described above is
incapable of maintaining independent existence, and in- found in all large rapid rockslides and beneath blockslides.
capable of verification, but there appear to be no other There are a variety of concepts of the role comminution
size requirements. Geometric logic dictates that the total may play in their motion. Many studies ignore the com-
surface area on all fragments is the original grain surface minution, assuming that it occurs before the landslide, or
area plus the twice the total area of cracks generated. Thus in the initial moments, and giving it no part in the mo-
grain numbers scale linearly with total crack area, and tion. In studies that consider comminution, opinions are
hence with the square of mean diameter (nominal grain divided as to whether it aids or hinders motion. Many
size). If grain fragments form without regard for size, it see it as an energy sink, whereas we see it as a temporary
follows that they will form randomly with the square of energy diversion. For certain, it is positive evidence for
grain size, and there will be more larger grains than resistance to motion, which we call friction, and proof that
smaller grains among the fragments, with a Power-law energy has been lost to friction, but is the loss to friction
gradient of +1.0, not 2.58. more, less, or the same as the loss that would occur if com-
If fragments are being formed by grain crushing, then minution had not occurred during the grain flow? Assum-
grains are being crushed between other grains. If they all ing that it occurred before leaves unanswered the ques-
are of similar strength, then those requiring less defor- tion of the means by which it occurred. Assuming that it
mation (ie smaller grains between larger grains) are pref- occurred in the initial moments requires an explanation
erentially crushed. Since the comminution process can- of where the energy came from to achieve it, and why the
not make grains any larger, continued comminution leads huge amount of elastic strain energy released in that in-
to more and more smaller grains relative to larger grains, stant is not all too abundantly evident? We know how rock
across all but the finest scales of grains that have been masses respond to sudden huge releases of elastic strain
created. The tail of diminishing grain numbers at the fin- energy. Where is the required ballistic ejecta sheet that
est end of the distribution will be continually pushed to has never been reported?
finer and finer sizes until grains reach the limit beyond Fragmentation of rock is of great industrial impor-
which they cannot maintain independent existence. tance, and has been much researched for some centuries,
How then, does a power-law gradient of 2.58 come to even before modern concepts of energy had been devel-
be so common? If generally no two adjacent grains of the oped. There are applications desiring only large frag-
same size are needed to fulfill the theoretical requirements ments, and others calling for intense fragmentation. What-
for a fractal dimension of 2.58, it cannot be because it ever the application, fragmentation is accomplished at a
gives all grains equal and minimum probability of being cost. It takes input of energy to fragment rock, and con-
further fractured as Sammis et al. (1987) believed. trolling the rate of input of energy can cause it to frag-
If we consider the fate of the fragments formed by grain ment in a specific, desired way. In industry, it costs money
crushing in force chains, we can see how the system can to apply the amount of energy that will break rock, and
self arrange to eliminate combinations of similar-sized so the concept of fragmentation energy arose; although
grains. When crushing-grain fragments are pushed from it is much researched; it is not well understood, and gen-
force chains, they are injected into lower pressure sur- erally is misunderstood. All methods for fragmenting rock
roundings. The lowest pressure surroundings are the pore use the same process a mass is deformed until it breaks;
spaces between grains, and so smaller grains are being if it is not deformed in any way, it will never break.
injected between larger grains at all scales, destroying Even in a granular mass, work is done to accomplish
combinations of similar-sized grains which require larger deformation; for the same amount of deformation, if the
void space same force is applied, the same work is done, whether or
As shown above, no particular fractal dimension not grains fragment. This work done is energy expended,
should be maintained once the modal grain size ap- not withstanding that the elastic component is recovered.
proaches the grinding limit; and so the significant in- After comminution, the energy expended can be divided
122 Mauri McSaveney Tim Davies

by the new surface area created by the rock breakage, and

8.3.2 Dynamic Fragmentation and Collapse
fracture surface energy is calculated this way. In indus-
trial applications, this is an energy loss to be paid for. But
the work done is the deformation of the granular mass Comminution is a major physical-weathering process that
(aside from energy that is lost to heat, which is lost creates smaller particles of rock from larger particles.
whether or not there is grain breakage). Deformation of Although the landslide source rock may sometimes be
the grain mass is an integral part of landslide motion, and described as intact, it contains many unhealed joint sys-
so is not an energy loss in landsliding. There is no actual tems, formed by a variety of stresses endured between
energy associated with the surface itself the energy is emplacement of the geological formation, and its expo-
expended in the process of creating new surfaces. Van der sure in outcrop. There are never landslides in completely
Waals forces are associated with proximity to molecules, intact rock. Some of the comminution is collapse of the
not proximity to surfaces. They are present internally, but original rock mass along such defects, with breakage
cannot be accessed. As granular solids are crushed finer caused by flexing or contact stresses during motion. In-
and finer, there is no net increase in the total energy of spection of finely fractured debris, however, shows that
the mass, other than any thermal energy from the fric- in addition to blocks defined by pre-existing joints, there
tion there is no energy trapped on the surface. Fracture also many fragments formed by breakage well below any
surface energy is not an energy loss in landslide motion, joint spacing (Fig. 8.7).
and is associated with the fracturing deformation and not Separation along pre-existing joints may occur at rela-
with the resulting surfaces. Its association with surfaces tively low stresses, depending on the degree of develop-
is through historical accident involving the measuring ment of the joints or secondary cementation, but break-
process, and many theoretical studies of its significance age of intact rock between joints requires high stresses,
are spurious. It does, however, have much useful applica- and hence occurs most commonly (but not exclusively)
tion to industry. To associate it with the surfaces is a mis- where high dynamic stresses develop in, for example,
interpretation of its significance; just as vehicle fuel con- large-volume, rapid, rockslides and those involving large
sumption is associated with the engine and the journey, fall height or high impact velocity. Breakage of unjointed
and not associated with the fuel, surface energy is associ- rock by dynamic mechanical stress is termed dynamic
ated with the material and its deformation (journey); it is fragmentation (Grady and Kipp 1987; McSaveney and
not associated with the surfaces. Davies 2006).
Fracture-surface-creation energy returns to grain flows To distinguish the high- and low-stress modes of com-
as grain kinetic energy and heat, and there is a major logi- minution of a parent body, Davies et al. (1999) used the
cal error in accounting for it as a separate entity. This logi- term collapse for separation along joints, and fragmenta-
cal error is perpetuated in recent literature (eg Locat et al. tion for the further comminution below that dictated by
2006), but had its origins in the 18th and 19th centuries joint spacing. In this context, collapse and fragmentation
(Belidor 1725; Rittinger 1867, cited in Bond 1952). A short have different effects on the behavior of the developing
quote from a historical summary of how rocks break granular mass. The distinction is useful in concept, but
(Bond 1955) reveals how easily the error arose in the trans- collapse and fragmentation are merely extreme end mem-
fer from industrial to scientific application: Rock break-
age is produced by deforming the rock, commonly under
pressure, until the resulting stress locally exceeds the break-
ing strength and crack tips form, usually on the surface.
The surrounding strain energy then flows to the new crack,
which is thereby extended to split the rock. When the rock
breaks, or the strain is otherwise released, the mechanical
energy input is transformed into heat. We now recognize
that cracking releases strain energy which flows (con-
centrates) away from the crack which is thereby extended.
When the strain is released, it is ultimately transformed
into heat, but the post-fracture transformation process
(deformation) is more relevant to deformation studies
than either the fracture process, or the heating. In blast-
ing and crushing, however. post-fracture deformation is
irrelevant, and the ultimate fate of released mechanical Fig. 8.7. Finely fractured debris in the Falling Mountain rock ava-
lanche showing breakage of grains to well below any pre-existing
energy is frictional heat. In landslides, the post-failure joint spacing. Field of view is about 2 m across, but the fractal na-
deformation is the landslide, but the ultimate fate of all ture of the grain-size distribution indicates that it would appear es-
the released mechanical energy is still frictional heating. sentially the same if viewed at any scale
Chapter 8 Rockslides and Their Motion 123

bers of a continuum of behavior of rock-mass defects in

8.3.3 Effect of Dispersive Stresses Caused by
the break-up of large rock masses. Our interest mostly is
Dynamic Grain Fragmentation
with the behavior of brittle rock masses at rapidly ap-
plied high stresses where the grains in the deforming mass
are broken and re-broken many times, so that no espe- Dynamic grain fragmentation always generates a high
cially weak defects remain, other than between grains. local pressure, but the proportion of the total grain mass
Since joint spacings are often fractal, the size distribu- being fragmented at any instant is often very small. For
tion of joint-bounded blocks is often fractal. Dunning example, the formation of a scree slope by intermittent
(2004) has shown that granular masses formed mostly rockfall results from and generates energetic breakage of
from collapse can not be distinguished from ones that grains in occasional, widely spaced events. Likewise, a
are much more intensely fragmented solely on the basis small rockslide may have very few grains simultaneously
of power-law gradient of the grain-number distribution fragmenting because most grains are free to slip or rotate
by grain size. in response to applied stresses, rather than being confined
The elastic-strain energy (W) released by failure of a and forced to deform and fragment. Spatially and tem-
unit volume of stressed rock is given by: porally widely separated dispersive stresses, even though
individually of high magnitude, need not pervasively af-
W = Q2/2E (8.8) fect the overall motion of the associated grain mass, and
thus may not result in noticeable morphological charac-
(Herget 1988), where Q is the rock strength and E is its teristics. A part of the reason for this is that grain masses
elasticity (Q and E assumed to be isotropic). It is avail- are very effective at absorbing and redistributing locally
able to do work only while it is being released. This is high stresses; this is why soldiers can be protected from
why it is absurd to assume that all, or most of the commi- individual high-velocity bullets by sandbags.
nution occurs in the initial moments of the landslide; sud- Where fragmentation events are more frequent and
den release of so much energy could not pass unnoticed. more closely-spaced, however, deposit morphologies be-
Nor is there the means to accumulate this much energy gin to exhibit characteristics indicative of enhanced mo-
in a static grain mass. bility during emplacement. This enhanced mobility comes
Note that the work achievable from the energy release from the added fluid pressure of fragmentation as il-
is that due to a single failure episode; the same grain mass lustrated in the Mohr-Coulomb relationship (Eq. 8.1). It
can repeatedly absorb and release the same amount of takes its most dramatic form in impact cratering when
energy each time that its increasing number of grains are 100% of the rock mass is fragmenting in an instant, but
stressed to failure (as distinct from relict tectonic or over- ~1 GPa pressures can dramatically alter mobility charac-
burden stresses that can only be released once). teristics by reducing frictional resistance when less than
In dynamic fragmentation, a grain begins to fail when 1% of grain mass is instantaneously fragmenting (1% of
its material Hugoniot elastic limit (HEL) is exceeded. The 1 GPa is equivalent to a piezometric head of 300 m of water
rate of progress to total failure depends on intrinsic ma- in the pore space).
terial properties, the rate of applied strain and the con-
fining stress. The static unconfined compressive strength
(UCS) of the material is the lowest possible HEL. For many 8.3.4 Fragmentation in Confined Layers
types of brittle rock in outcrop, the HEL is of the order of
1 GPa (Melosh 1997). It has been stated (Knoeberl 1997) We first consider rockslides in which rock is fragmented
that The only known process that produces shock pres- in a relatively thin layer beneath a large intact rock mass.
sures exceeding the HELs of most crustal rocks and min- This situation is very similar to that frequently investi-
erals in nature is impact cratering. Volcanic processes are gated in laboratory tests. Common landslides of this type
not known to exceed 0.5 to 1 GPa. But such thinking arises are long-runout, low-angle block slides in which low slid-
through misunderstanding the HEL, which includes, but ing friction between rock surfaces is necessary to explain
is not restricted to, shock compression. Although the observations.
shock fronts associated with the formation and failure of Shear banding is commonly reported in blocksliding
force chains in rockslides are in the lower range of what (e.g., Anders et al. 2000) and some volcanic debris ava-
might commonly be perceived as shock compression, it lanches (Clavero et al. 2002), and less commonly in non-
is nevertheless HEL exceedence. In the moment of grain volcanic debris (rock) avalanches (Yarnold and Lombard
failure, the crushing grain should not be described as a 1989). In our experience, shear bands are not absent in
solid, above their HELs all materials are high-pressure non-volcanic debris avalanches, but they are not visually
vapours, and the crushing grain effectively is a high-pres- dominant.
sure pore fluid in the moment of crushing a very ephem- Consider a thin, granular, shearing, fragmenting stra-
eral fluid. tum between a moving non-deforming layer and a sta-
124 Mauri McSaveney Tim Davies

Hence in Eq. 8.11, 0.01 = k 2 (0.75 0.3) 10 6 , so

k2 = 2 108.
From Eqs. 8.9, 8.10 and 8.11,

PF = k1 k2(POB PC)/(1 + k1 k2) (8.12)

k1 k2 = 109 2 108 = 20 , so

PF = 20(POB PC)/21 = 0.95(POB PC) (8.13)


PE = POB PF = 0.05 POB + 0.95 PC (8.14)

Fig. 8.8. Definition sketch of stresses operating in a thin, granular,

In terrestrial landslides POB may be up to ~7 MPa
shearing, fragmenting stratum between a moving non-deforming (200300 m deep), while PC is ~0.2 MPa, in which (ex-
layer and a stationary base treme) case PE = 0.35 + 0.19 = 0.37 MPa ~ 2PC, so:

tionary base (Fig. 8.8): in which POB is the overburden PE < 2PC O(PC) (8.15)
pressure at top of the fragmenting layer, PF is the average
fragmentation pressure in the fragmenting stratum and and the steady-state or time-averaged effective stress in
PE is the effective intergranular direct stress in it. Stresses the fragmenting layer of a landslide is of the order of the
and strain rate are assumed time-invariant. minimum stress needed to maintain fragmentation.
Applying a vertical stress balance at the top of the frag-
menting stratum:
8.4 Block Slides
PE = POB PF (8.9)
We apply this theory first to blockslides, which obviously
Each fragmenting grain generates a pressure = Q, so conform closely to our model of confined granular flow.

PF = k1Ff (8.10)
8.4.1 The Problem
where Ff = proportion of simultaneously fragmenting
grain mass; k1 = Q ~ 109 Pa. Blockslides are relatively intact blocks that lie some dis-
The simultaneously fragmenting proportion (at a given tance from their original locations, having moved on
shear rate) will depend on the amount by which the ef- gentle slopes. Well-known examples are the Heart Moun-
fective intergranular direct stress PE exceeds the minimum tain slide, Montana and Wyoming, USA (Prostka 1978);
needed to cause fragmentation PC: the Horse Creek and South Creek slides, Idaho, USA
(Beutner 1972), the Bearpaw Mountains slide, Montana
Ff = k2(PE PC) (8.11) (Gucwa and Kehle 1978) and the Dakota Group rockslides,
Colorado, USA (Braddock 1978). These have volumes of
assuming linear dependency, which seems reasonable. tens to hundreds of cubic kilometers, thicknesses of about
In a simulation of the Falling Mountain non-volcanic a kilometer, and moved typically tens of kilometers on
debris avalanche (Davies and McSaveney 2002), the re- slopes as gentle as 2. How do they move so far on such
quired mean fragmentation stress explaining the runout low-angle slopes? Is their motion fast or slow?
distance was ~2.5 MPa, implying Ff ~ 0.01. This simula-
tion simply used the longitudinal component of disper-
sive pressure to provide the force needed to generate the 8.4.2 Our Approach to a Solution
observed runout, without affecting effective stress, so that
POB = PE. The mean overburden pressure resulting in this New Zealands prehistoric 2.1 km3 Waikaremoana land-
stress was ~0.75 MPa. The minimum stress causing slide, (Read et al. 1991) is a very useful blockslide example.
fragmentation (PC) was that at the base of the relatively About two thousand years ago it dammed Waikaretaheke
unfragmented 10-m deep carapace, ~0.3 MPa. This ob- River to form Lake Waikaremoana, a large lake that today
servation suggests that PC ~ 103 Unconfined Compres- forms the centrepiece of Urewera National Park. It has
sive Strength. both blockslide and rock-avalanche components. Follow-
Chapter 8 Rockslides and Their Motion 125

ing 2 km of displacement on a 5.58 slope, its 1.4 km3 the ratio of inertial to gravitational forces in the model was
blockslide portion ploughed into rock-avalanche debris, the same as in the landslide, so that the ability of the kinetic
raising a 150 m high debris mound (Fig. 8.1). Using an en- energy or momentum in the model block to displace granu-
ergy-balance approach and assuming 80% efficiency, lar material would be similar to that in the field.
Beetham (1984) estimated an impact speed of 26 m s1. We satisfied the criterion by scaling the block and grain
Davies et al. (2006) sought confirmation of this assump- masses in proportion to the cube of the linear scale, by
tion through a small physical model. When the model indi- using sand of the same bulk density (about 2 t m3) and
cated an even higher impact speed, and drilling showed internal friction coefficient (about 0.6) as found in the
highly comminuted rock on the slide plane, we interpreted prototype, and by adjusting the model block weight to
the remarkably low coefficient of friction in terms of the correspond to that of the Waikaremoana block. Since the
ephemeral fluid pressure due to dynamic fragmentation. acceleration due to gravity is identical in model and pro-
totype, the two situations are dynamically similar. With
this similarity of forces, the velocity scale is equal to the
8.5 The Waikaremoana Blockslide Model square root of the linear scale (Yalin 1971) in this case
(1 : 10 000)0.5 = 1 : 100.
8.5.1 Preparing the Physical Model The model apparatus (Fig. 8.9) conceptually repre-
sented only a small part of the total Waikaremoana land-
To assess how fast the Waikaremoana block slid to push slide, specifically where the block stopped against rock-ava-
up the mound, we built a small (1 : 10 000 linear scale) lanche debris at the edge of a proto-valley, and the mound
laboratory model. The crucial design criterion was that was formed. It was not required to model any other land-
slide part. The model block was swung from four suspen-
sion wires to hit the grain mass (sand) in the correct atti-
tude at a range of speeds, and we observed the reposition-
ing of the sand using a high-speed video camera (200 fps)
which allowed us to measure block and sand speeds.

8.5.2 Model Results

To achieve a satisfactory model mound, two conditions

were needed. First, the sand needed to move with the
block as they both impacted the proto-valley side. We
Fig. 8.9. Diagrammatic sketch of the model used to determine the
achieved this by putting the sand some distance from the
speed of the Waikaremoana blockslide at the moment of impact with barrier, and allowed the block to swing into it faster than
the opposing valley wall required; it then set the sand in motion, so that both were

Fig. 8.10.
Prototype and model debris
126 Mauri McSaveney Tim Davies

Fig. 8.11.
Shaded relief map of the Waika-
remoana landslide at Lake Wai-
karemoana, showing the rela-
tionship between the blockslide
and rock avalanche components

moving at a lower velocity when they hit the barrier. Sec- To achieve this acceleration, basal and lateral friction co-
ond, the block and the sand needed to hit the barrier at efficients must have been much lower than normally as-
about 0.4 m s1. sociated with rock-on-rock sliding (0.6). It is inconceiv-
These conditions produced a mound about 15 mm high, able that an earthquake gave the block a high initial speed
correctly representing the 150 m height in the field. As well, and its 40 m s1 impact was reached after decelerating
the morphology of the mound, with pressure ridges, was under normal friction. This requires an impossible ini-
strikingly similar to that in the field (Figs. 8.10 and 8.11). tial speed of ~600 m s1. We inferred that the intact block
accelerated to a maximum 40 m s1 during its 2-km travel
while sliding on the 5.58 basal surface and its left-lateral
8.5.3 Applying the Model Results to the Field margin (the right-lateral margin was unconfined). There
has to have been very low frictional resistance to motion
The model correctly scaled the forces during the impact, of the block. If the block started from rest, an average
and so valid conclusions can be drawn regarding the field acceleration of 0.4 m s2 is needed, requiring a mean fric-
processes. We concluded that the Waikaremoana block- tion coefficient of 0.06 about 1/10th of normal friction.
slide was rapid, with an impact speed of ~40 m s1. It could A number of possibilities, including high pore-water
not emplace the mound with an impact speed of only pressures and undrained loading of a saturated substra-
26 m s1. The 33% higher speed indicates a less efficient tum, were investigated for explaining this low friction and
conversion of kinetic energy to potential energy in the eliminated for a variety of reasons (Beetham 1983; Read
collision than the previously assumed 80%. The model et al. 1991; Davies et al. 2006). Rock fragmentation at the
also showed that the deformed debris mass was not at sliding interface was inevitable, and a thin (300 mm) layer
rest at the time of impact; it had to have been in motion of finely-ground rock at the base of the displaced block
to form a mound with pressure ridges. This demon- was indicated by drilling through this horizon (Davies
strated that the rock avalanche was generated by collapse et al. 2006). This was interpreted as fragmented sandstone.
and disintegration of the front of the block and that the A high ephemeral fluid pressure, generated by fragment-
two landslide parts were synchronous. ing clasts, can readily substitute for pore-water pressure
The block traveled 2 km before stopping against a and provide sufficient vertical force to support much of
gorge wall and debris. To reach 40 m s1 in this distance the block weight during motion, and hence reduce the
requires an average down-slope acceleration of 0.4 m s2. frictional resistance to motion.
Chapter 8 Rockslides and Their Motion 127

The rock above and below the fragmenting material In material-failure theory, all failure utilizes pre-exist-
also are available to fragment, and so the system cannot ing defects. As stress grows in a body, suitably oriented
run out of pore fluid until it runs out of fragmentation- pre-existing defects grow and coalesce, until a through-
inducing motion. going defect is formed to permit failure. When imper-
The static unconfined compressive strength of the fectly elastic rock masses are deformed by folding, some
sandstone at Waikaremoana is about 50 MPa (Read 1979), slip between some beds is a kinematic requirement, and
so its dynamic Q may be about 100200 MPa (or even so bedding-plane shears develop in folded rocks. These
greater at the ~100 s1 shear rate required for a velocity are often exploited in landslides. At Waikaremoana, the
differential of ~30 m s1 to take place across 0.3 m or less). axis of shortening due to folding is not easy to define with
Applying Eq. 8.14, the frictional retarding stress in shear any precision because the folding is complicated, and it
bands should be ~0.95 0.2 + 0.05 6.6 = 0.52 MPa. The may result from an aggregation of episodic deformation
geostatic stress due to the weight of the 275-m deep block events from slip on a number of different faults. Each sepa-
is ~6.6 MPa, so to support most of this requires about one rate event may have its own orientation of bedding-plane
part in ~30 of the participating grain mass to be frag- shear. The gross pattern of folding would not have pro-
menting at any time to support the weight of the block. If duced bedding-plane shear parallel to the direction of
each force chain across the width of a shear band con- eventual landslide motion.
tains about 10 grains, then about 1 part in 3 103 of a A clue to how a failure surface may have developed at
100 mm thick shearing layer needs to be fragmenting si- 275 m depth beneath the Waikaremoana landslide per-
multaneously. The net downslope stress (gravity compo- haps can be found elsewhere in the local structural geol-
nent minus friction) is then 6.6sin(7) 0.52 = 0.28 MPa. ogy. The SE flanks of the adjacent hills are dip slopes. In
The block mass per unit area is 0.69 kt m2, so the downs- the vicinity of the headscarp of the landslide, the dip slope
lope stress would cause block acceleration of 0.4 m s2. averages 19.1 while beyond the blockslide toe it averages
Two kilometers of travel at this acceleration would de- 14.7. These correspond closely to nearby measured dips
velop a velocity of about 40 m s1. on exposed bedrock and are probably more precise esti-
mates of the bedding attitude than direct measurements
at small outcrops. They indicate a twist in the bedding in
8.5.4 A Fragmentation Model for Initiation the vicinity of the landslide. The left lateral margin of the
of the Waikaremoana Landslide Waikaremoana blockslide is inferred by Beetham (1983) to
follow a fault. It is normal to the expected direction of prin-
Davies et al. (2006) showed that fragmentation in a con- cipal tension from the twist motion, however, and we could
fined layer could explain the high-speed emplacement infer that the left lateral release surface for the blockslide
of the blockslide with low frictional resistance, but we might be a tension joint formed by the twist folding.
did not explain how the blockslide could develop the Twist-derived tensional jointing offers a simple expla-
initial shear deformation rate to induce widespread nation for that major release surface, but the same twist
fragmentation at its base. There is reason to believe that deformation offers several possible explanations for the
it started during strong ground shaking in an earthquake other major release surface, the basal slide plane. First,
(Beetham 1983). It was once thought to have slid on the the direction of bedding-plane slip in the twist is parallel
interface between mudstone and sandstone strata to the direction of eventual slip of the blockslide.
(Beetham 1983), but drilling showed it to have failed The local bedrock involved in the landslide is a strongly
entirely in sandstone (Davies et al. 2006). Although cemented calcareous sandstone. Its behavior under im-
part of the failure surface is thought to have been a bed- pulse loading could be highly elastic, especially in the
ding plane, it cannot all have been a bedding plane, be- near-surface environment. When a thin elastic sheet is
cause the bedding-plane geometry is impossible for fail- twisted, it has two stable modes: in one, the center of the
ure to occur that way (Beetham 1983; Riley and Read sheet is down; and in the other, it is up. If it is done ex-
1992). The failure surface had to climb up through the perimentally, a small sheet can be flipped from one to the
stratigraphic pile before slip could occur. One block of other mode with little effort being required. At field scale,
the landslide is essentially intact, and is some 3.5 km long. the up mode is unlikely to be adopted at depth because
If it was differentially flexed in its 2 km journey along it requires formation of void space beneath the elastic
the failure surface, flexure can only have been very mi- sheet, but for surface strata, the up mode is a transient
nor, and flexed over a rise (opening joints near the sur- option under impulse loading. Perhaps, the hard sand-
face) and not through a hollow (opening joints near the stone that forms the crest and dip slope of the range was
base and compressing the surface). Hence, we infer that coseismically twisted some 2200 years ago, and the block
there is little curvature on the failure surface in the di- at the SW end of the range popped up, with a ground ac-
rection of sliding. We can reasonably infer that it was celeration normal to the bedding plane of >1g, while be-
like this from the start. low it, the rock mass popped down. With rock masses
128 Mauri McSaveney Tim Davies

accelerating rapidly in opposite directions, a zone of rock vicinity of the headscarp, the fallback may have been trivial
mass would be put in tension, a myriad of favourably ori- and unable to destabilize the rock mass beyond there.
ented defects would coalesce, and a surface of separation The total twist is ~4.4 (0.0775 radians). If this
would form in tensional failure. This offers one potential has occurred over ~5 million years, the annual rate is
explanation of the separation. ~1.55 108 radians per year, or ~77.5 rad per event, as-
Another explanation also uses the twist, but in combi- suming a recurrence interval of perhaps 5000 years for
nation with the local geomorphology. The range is a the coseismic twisting (we have no data on this). Over
cuesta (Fig. 8.11), with one flank covered by a thick the 1500-m width of the Waikaremoana blockslide, the
strongly cemented sandstone. During coseismic twisting widest gap(s) would be ~116 mm. The geometry is shown
of the underlying rock mass, a portion of the relatively diagrammatically in Fig. 8.12. The velocity, v, of the outer
thin skin, in close proximity to two free edges of the beds, edge of the block as it pivots back in free fall under
over the twist may have failed to fold (due to inertia) as gravity (g) is given by v2 = 2gs = 1.5 m s1. A 1-m2 column
the rock underlying it folded. A zone at the base of the of rock, 275 m high, of density 2.5 t m3, would have kinetic
strong mass was put in tension, a myriad of favourably energy of 0.784 MJ after a freefall of 116 mm under gravity.
oriented defects (pre-existing bedding-plane shears) coa- Either of these alternatives would have left 2.1 km3 of
lesced, and a surface of separation formed in tensional rock mass literally up in the air (though possibly above a
failure, delaminating a slab of strong rock from the un- vacuum). We might suppose that the tensional failure was
derlying beds. This could leave a rock slab a little larger not a single clean break, but a multitude of propagating
than 2.1 km3 unsupported above a double-wedge-shaped cracks, so that many small pieces of rock were torn be-
gap. The slab then could fracture along pre-existing planar tween the two parted surfaces. In addition, the mass would
defects (an orthogonal set of tensional joints from earlier not fall directly back on its pre-failure position. Thus, the
folding episodes. The portion of this rock mass towards rock mass would fall back with much stress concentra-
the lower-dip end of the twist (the SE end of the range, to- tion at asperities, which could be instantly crushed. In
wards the proto-gorge) would be raised highest above its addition to removing asperities from the failure surface,
substrate and fall furthest on failure. The delaminating gap however, the instantaneous crushing would also remove
would decrease progressively NE along the range. In the most frictional resistance to downslope motion. The frag-
menting rock would behave as a heavy vapor at a pres-
sure of the Hugoniot elastic limit (HEL) of the rock un-
der the local environmental conditions (of confining pres-
sure, temperature, and rock strain rate). The HEL of the
cemented calcareous sandstone is greater than the static
unconfined compressive strength (~50 MPa) and prob-
ably in the region of 100200 MPa, whereas the overbur-
den pressure beneath the blockslide at Waikaremoana
was ~6.6 MPa. Hence we can suppose that the block pro-
ceeded to rapidly accelerate downslope at ~9.81sin(7.5)
= ~1.28 m s2, initially essentially with no frictional re-
sistance. The fragmentation process, however, is self-
regulating: if there is no effective normal load in the frag-
menting layer, the grains will not be stressed, and so will
not continue to fragment. A balance should be quickly
achieved in which the load carried by the ephemeral pres-
sure of fragmenting grains leaves sufficient load to main-
tain some fragmenting grains. We surmise that fragmen-
tation of rock at the base of the block was initially close
to 100% of all rock particles, but quickly dropped to
~0.03% of the shearing layer fragmenting at any instant,
and continued at that level until block motion abruptly
stopped when it hit the opposite valley wall. In order to hit
the wall at ~40 m s1, an average acceleration of 0.3 m s2
was required, so it is not likely that an initial downslope
acceleration of 1.28 m s2 was maintained more than
Fig. 8.12. Our concept of how the Waikaremoana landslide may have
momentarily. What is more important than the duration
been initiated as an elastic response to coseismic folding of near- of high acceleration is that the mass may have had its high-
surface strata est downslope acceleration when starting to slide, de-
Chapter 8 Rockslides and Their Motion 129

clining to some steady-state value as the mass gained ve- and Petley 2003), but it seems possible that rock fragmen-
locity. In this model, the initial lowest friction was caused tation may have played some role there too.
by massive fragmentation of rock at the base of the block Some of the anomalously low travel angles exhibited
as it fell back in a 1-g freefall from being raised (or left) by rock-avalanche deposits (e.g., Cruden 1980) can be
into the air, being a near-surface elastic response of a rock explained by the notion that some rock avalanches move
mass to coseismic folding. Although it appears to be a initially as blockslides. An initial blockslide phase at low
feasible model, we do not know of a means to test whether friction allows faster acceleration than under normal fric-
it occurred. tional resistance. But not all rock avalanches need have
an initial blockslide phase of any significant duration.

8.5.5 Other Rockslide Phenomena

8.5.7 Dry Rock Avalanches
We now consider rockslides involving fragmentation
throughout most the moving debris mass, other than Rock avalanches generally move as dry granular flows,
within the upper few meters where low confining stresses because the large volume (~20%) of void space created
readily permit inter-granular slip and rotation. Large rock by the observed pervasive fragmentation of initially mas-
avalanches have a fascinating morphological character- sive rock is several orders of magnitude more than suffi-
istic; their runout lengths are proportionally greater for cient to prevent saturation of the debris mass by any wa-
their volume than those of their smaller counterparts, to ter present in the parent rock or along the runout path
a degree that increases markedly with the volume of the
deposit. This is the so-called size effect (e.g., Hs 1975,
1978; Scheidegger 1973). We have investigated this
anomaly (McSaveney 1978; Davies 1982; Davies and
McSaveney 1999, 2002, 2003; Davies et al. 1999; McSaveney
and Davies 2002, 2003; McSaveney et al. 2002), and con-
clude that, rather than requiring some novel physical
mechanism (e.g., Shreve 1968; Goguel 1978; Melosh 1979;
Campbell 1989; Campbell et al. 1995; Collins and Melosh
2003), or extensive occurrence of some particular sub-
strate condition (e.g., Sassa et al. 2004; Abele 1997; Hungr
and Evans 2004) to generate low basal friction beneath
the mass of translating debris, the continuing process of
dynamic fragmentation during the runout generates suf-
ficient internal fluid pressure that, under conditions of
normal intergranular and basal friction, the debris can
spread to a greater extent than would occur without frag-
mentation. This extra spreading results in the extended
distal runout observed in rock-avalanche deposits, and is
accompanied by reduced runout of proximal material.

8.5.6 Transition from Blockslide to Rock Avalanche

We have shown above that dynamic-fragmentation is ca-

pable of explaining the extraordinarily rapid acceleration
needed in some blockslides to cause their subsequent
motion and effects. The Avalanche Lake rock avalanche
in Canada, began as a blockslide, and apparently needs a
basal friction coefficient of 0.04 to explain its motion
(Hungr 1995). The acceleration of the initial blocks of the
1980 Mt. St. Helens debris avalanche, computed from the
Rosenquist series of photographs, requires a basal fric-
Fig. 8.13. Our concept of how the Falling Mountain rock avalanche
tion coefficient of 0.1 (Voight 1981). The very rapid mo- may have initiated as a rockslide. The cornflake analogy is because
tion of the failed blocks of Mt. Toc in the Vaiont disaster force chains are what often prevent breakfast cereals from pouring
of 1963 has been attributed to low-strength clay (Kilburn from the box, requiring shaking to disrupt the force chains
130 Mauri McSaveney Tim Davies

(McSaveney and Davies in press). They are, however, We do not know the appropriate dynamic strength at
highly erosive, tending to incorporate any readily deform- the local strain rates (>10 s1) and overburden pressures
able substrates encountered along their paths and to de- (>0.3 MPa) causing fragmentation during the rock ava-
posit them in the distal bases of their deposits this can lanche but estimate that it is likely to be at least ~500 MPa
include rivers, lakes and groundwater systems. (following Kobayashi 1970). Our previous simulation
Our concept of grain conditions in rock-avalanche (Davies and McSaveney 2002) showed that the avalanche
motion is illustrated in Fig. 8.13; the stars represent grains runout could be represented by adding an isotropic dis-
of various sizes fragmenting simultaneously. persive stress equivalent to 0 5% of the grain mass frag-
menting simultaneously in a 1-dimensional simulation
(DAN (Hungr 1995)). We now modify this by first calcu-
8.5.8 Falling Mountain lating the resisting stress in the shearing layer, which is
equal to the basal resisting stress. Since shear takes place
In the case of the Falling Mountain deposit (Fig. 8.2) in- in localized bands, this is given by Eq. 8.15, and is
vestigated by Davies and McSaveney (2002), the rock is 0.34 MPa. This results in an apparent friction angle of 20.
a highly indurated greywacke sandstone with a static, The depth-averaged overburden pressure resisted by frag-
unconfined, intact compressive strength of ~280 MPa. mentations to leave this residual shear resistance is

Fig. 8.14.
Aerial view of the small prehis-
toric South Ashburton rockslide
which fell onto and eroded a wet,
readily mobilized fine substrate,
but which does not exhibit long
runout (see McSaveney et al.
2000; McSaveney and Davies
2005, 2006)
Chapter 8 Rockslides and Their Motion 131

0.5 MPa, and the spatial density of simultaneous fragmen-

8.5.9 South Ashburton Landslide
tations needed to achieve this is about 0.01%. Rerunning
the DAN simulation model (of Davies and McSaveney
2002) with a friction angle of 20 and an isotropic frag- New Zealands South Ashburton landslide (Fig. 8.14;
mentation-induced dispersive pressure of 0.5 MPa yields McSaveney et al. 2000) is an unusual and instruc
exactly the correct runout of the avalanche. tive rockslide; it fell onto, and ran out over a fine-grained,
Our previous simulation used trial-and-error to ob- saturated, and deformable substrate, but it does not
tain the correct runout by varying the spatial density of exhibit classical long runout (McSaveney and Davies
simultaneous fragmentations; the present simulation was 2005, 2006). Because the landslide was too small to
based on calculated internal friction resulting from utilize dynamic fragmentation to any significant de-
fractally distributed, isotropically oriented shear bands gree, its runout could be modeled with dynamic simi-
below a passive carapace, and their effect on internal fric- larity using sand (McSaveney et al. 2000). Our physi-
tion calculated by Eq. 8.13. The success of the present cal modeling of it (Fig. 8.15) suggested that it fell first
simulation independent of any calibration is remarkable. as a block, which then disintegrated and spread at the
If all the fragmentation observed in the rock-avalanche foot of the slope.
deposit were to occur at the start of the motion, as often
has been suggested, the concentration of GPa-range forces
acting simultaneously would cause an instantaneous ex- 8.6 Conclusion
plosive fragmentation of the whole mass, and its disper-
sion as a (probably mushroom-shaped) dust cloud, with The motion of large rockslides is most readily un-
no evident translational mass movement this does not derstood as being controlled by comminuting-grain-
occur. Equation 8.8 suggests that complete fragmentation flow dynamics, where grains crushing under large
of the 55 106 m3 volume of the Falling Mountain rock dynamic stresses in force chains form an ephemeral
avalanche in 1 second at the start of its motion would have high-pressure fluid in the pore space between larger
released elastic-strain energy equivalent to a nuclear ex- grains at all scales and reduce the overall resistance
plosion in the several-megaton range, if complete frag- to motion in proportion to the fraction of the grain
mentation took place in 1000 stages. But the energy for mass that is being crushed at any instant. The motion
fragmentation is fed from the kinetic energy of shear- of small rockslides where grain comminution is un-
strain deformation, and there obviously is insufficient important is controlled by grain-to-grain friction and
kinetic energy available initially to cause such an explo- intergranular stresses determined by the non-frag-
sion at the start of the landslide. The concept of near- menting grains and any other pore fluid which may
instantaneous, initial fragmentation is thus an absurdity. be present.

Fig. 8.15.
Prototype and physical model of
the South Ashburton rockslide.
The most satisfactory model was
with the debris mass gaining
momentum in a container, be-
fore flowing onto a high-friction
glass plate coated with a thin
layer of loose sand forming an
erodible substrate. This eroded
substrate was carried at the base
of the rockslide front and formed
the raised distal rim. Without
the erodible substrate, the raised
rim could not be formed. Simi-
larity of model and prototype
indicate that the prototype flowed
with normal intergranular fric-
tion similar to that of the sand
132 Mauri McSaveney Tim Davies

Crosta GB, Frattini P, Fusi N (2007) Fragmentation in the Val Pola

Acknowledgment rock avalanche, Italian Alps. J Geophys Res (in press)
Cruden DM (1980) The anatomy of landslides. Can Geotech J 17:
This research has been supported by the New Zealand Cruden DM, Varnes DJ (1996) Landslide types and processes. In:
Foundation for Research, Science and Technology. Turner AK, Schuster RL (eds) Landslides: investigation and miti-
Alexander Strom, Stuart Dunning, and Giovanni Crosta gation. Special Report 247, Transportation research board, US
helped us develop our ideas on particle-size distributions National research council, Washington, D.C., pp 3675
Davies TRH (1982) Spreading of rock avalanche debris by mechani-
in comminuting grain flows. We gratefully acknowledge
cal fluidisation. Rock Mech 15:924
many other colleagues, both believers and doubters, whose Davies TRH, McSaveney MJ (1999) Runout of dry granular ava-
discussions have helped mature our concepts. lanches. Can Geotech J 36(2):313320
Davies TRH, McSaveney MJ (2002) Dynamic simulation of the mo-
tion of fragmenting rock avalanches. Can Geotech J 39:789798
References Davies TRH, McSaveney MJ (2006) Runout of rock avalanches and
volcanic debris avalanche. In: Picarelli L (ed) Proceedings of the
Abele G (1997) Rockslide movement supported by the mobilization International Conference on fast slope movements: prediction,
of groundwater-saturated valley floor sediments. Z Geomorphol and prevention for risk mitigation. Naples, May 1113, 2003, vol. 2
41(1):120 Davies TRH, McSaveney MJ, Hodgson KA (1999) A fragmentation-
Allegre CJ, Mouel JLL, Provost A (1982) Scaling rules in rock frac- spreading model for long-runout rock avalanches. Can Geotech
ture and possible implications for earthquake prediction. Nature J 36:10961110
297:4749 Davies TRH, McSaveney MJ, Beetham RD (2006) Rapid block glides
An LJ, Sammis CG (1994) Particle size distribution of cataclastic fault slide-surface fragmentation in New Zealands Waikaremoana
materials from Southern California: a 3-D study. Pure Appl Geo- landslide. Q J Eng Geol Hydroge 39:115129
phys 143:203227 Dunning SA (2004) Rock avalanches in high mountains A sedimen-
Anders MH, Aharonov E, Walsh JJ (2000) Stratified granular media tological approach. University of Luton, unpublished PhD Thesis
beneath large slide blocks; implications for mode of emplace- Evans SG, Hungr O, Enegren EG (1994) The Avalanche Lake rock
ment. Geology 28:971974 avalanche, Mackenzie Mountains, Northwest Territories, Canada:
Anthony JL, Marone C (2005) Influence of particle characteristics description, dating and dynamics. Can Geotech J 31:749768
on granular friction. J Geophys Res 110: B08409 Doi: 10.1029/ Goguel J (1978) Scale-dependent rockslide mechanisms, with em-
2004 JB003399 phasis on the role of pore-fluid vaporization. In: Voight B (ed)
Beetham RD (1983) Seismicity and landsliding with special atten- Rockslides and avalanches, I, : developments in geotechnical en-
tion to New Zealand. Unpubl. MSc thesis, University of London gineering. 14A, Elsevier, Amsterdam, pp 693705
Belidor BF (1725) Nouveau cours de mathmatique lusage Grady DE, Kipp ME (1987) Dynamic rock fragmentation. In: Atkinson
dartillerie et du gnie. Paris, 505 p BK (ed) Fracture mechanics of rock. Academic Press, London,
Belousov A, Belousova M, Voight B (1999) Multiple edifice failures, pp 429475
debris avalanches and associated eruptions in the Holocene history Gucwa PR, Kehle RO (1978) Bearpaw Mountains Rockslide, Montana,
of Shiveluch Volcano, Kamchatka, Russia. B Volcanol 6:324342 U.S.A. In: Voight B (ed) Rockslides and avalanches, I: developments
Beutner EC (1972) Reverse gravitative movement on earlier over- in geotechnical engineering. 14A, Elsevier, Amsterdam, pp 393421
thrusts, Lemhi Range, Idaho. Geological Society of America Bul- Hazzard JF, Mair K (2003) The importance of the third dimension in
letin 83:839846 granular shear. Geophsy Res Lett 30, doi: 10.1029/2003GL017534
Beutner EC, Gerbi GB (2006) Catastrophic emplacement of the Heart Herget G (1988) Stresses in rock. Balkema, Rotterdam, 179 p
Mountain blockslide, Wyoming and Montana, USA. GSA Bull Howell DW, Behringer RP, Veje CT (1999a) Fluctuations in granular
117:724735; doi: 10.1130/B25451.1 media. Chaos 9:559572
Bond FC (1952) The third theory of comminution. T Am I Min Met Howell DW, Behringer RP, Veje CT (1999b) Stress fluctuations in a
Eng 193:494496 2D granular Couette experiment: a continuous transition. Phys
Bond FC (1955) How does rock break? The Scientific Monthly 81: Rev Lett 26:52415244
196198 Hs KJ (1975) Catastrophic debris streams (Sturzstroms) generated
Braddock WA (1978) Dakota Group rockslides, northern Front Range, by rockfalls. Geol Soc Am Bull 86:123140
Colorado, U.S.A. In: Voight B (ed) Rockslides and avalanches, I. Hungr O (1995) A model for the runout analysis of rapid flow slides,
Developments in Geotechnical Engineering, 14A, Elsevier, debris flows and avalanches. Can Geotech J 32(4):610623
Amsterdam, pp 439479 Hungr O, Evans SG (2004) Entrainment of debris in rock avalanches:
Campbell CS (1989) Self-lubrication for long-runout landslides. an analysis of a long run-out mechanism. Geol Soc Am Bull
J Geol 97:653665 116:12401252
Campbell CS (2006) Granular material flows an overview. Powder Kilburn CRJ, Petley DN (2003) Forecasting giant, catastrophic slope
Technol 162:208229 collapse: lessons from Vajont, Northern Italy. Geomorphology
Campbell CS, Brennen CE (1985) Computer simulation of granular 54(12):21-32
shear flows. J Fluid Mech 151:167188 Knoeberl C (1997) Impact cratering: an overview of mineralogical
Clavero JE, Sparks RJS, Huppert HE, Dade WB (2002) Geological and geochemical aspects http://www/univie.ac.at/geochemistry/
constraints on the emplacement mechanism of the Parincota impp.html
debris avalanche, northern Chile. B Volcanol 64: 4054. doi: Kobayashi R (1970) On mechanical behaviours of rocks under vari-
10.1007/s0044500101830 ous loading rates. Rock Mech Jpn, 1:5658
Collins GS, Melosh HJ (2003) Acoustic fluidization and the extraor- Liu CH, Nagel SR, Schecter DA, Coppersmith SN, Majumdar S,
dinary mobility of Sturzstroms. J Geophys Res 108(B10): 2473, Narayan O, Witten TA (1995) Force fluctuations in bead packs.
doi:10.1029/2003JB002465 Science 269:513515
Chapter 8 Rockslides and Their Motion 133

Locat P, Couture R, Leroueil S, Locat J, Jaboyedoff M (2006) Frag- Riley PB, Read SAL (1992) Lake Waikaremoana: present day stabil-
mentation energy in rock avalanches. Can Geotech J 43:830851 ity of landslide barrier. In: Bell DH (ed) Landslides : Proceedings
Mair K, Frye KM, Marone C (2001) Influence of grain characteristics of the sixth International symposium 1014 February 1992,
on the friction of granular shear zones. J Geophys Res 107: 2219 Christchurch, Balkema, Rotterdam, pp 12491256
Doi:10.1029/2001BJ000516 Sammis C, King G, Biegel R (1987) The kinematics of gouge forma-
McSaveney MJ (1978) Sherman Glacier rock avalanche, Alaska, USA. tion. Pure Appl Geophys 125:777812
In: Voight B (ed) Rockslides and Avalanches 1, Natural Phenom- Sassa K, Fukuoka H, Wang G, Ishikawa N (2004) Undrained dynamic-
ena. Developments in Geotechnical Engineering 14A(6):197258 loading ring-shear apparatus and its application to landslide
McSaveney MJ (2002) Recent rockfalls and rock avalanches in Mount dynamics. Landslides 1:719
Cook National Park, New Zealand. In: Evans SG, DeGraff JV (eds) Cat- Scheideigger AE (1973) On the prediction of the reach and velocity
astrophic landslides: occurrence, mechanisms and mobility. Geologi- of catastrophic landslides. Rock Mech 5:231236
cal Society of America Reviews in Engineering Geology 15:3570 Sheve RL (1968) The Blackhawk landslide. Geological society of
McSaveney MJ, Davies TRH, Hodgson KA (2000) A contrast in deposit America, Special paper, 108, 47 pp
style and process between large and small rock avalanches. In: Turcotte DL (1986) Fractals and fragmentation. J Geophys Res 91:
Bromhead E, Dixon D, Ibsen ML (eds) Landslides in research, theory 19211926
and practice. Thomas Telford Publishing, London, pp 10531058 Varnes DJ (1978) Slope movement types and processes. In: Schuster
McSaveney MJ, Davies TRH, McSaveney MJ (2006) Rapid rock-mass RL, Krizek RJ (eds) Special report 176: Landslides: analysis and
flow with dynamic fragmentation. In: Evans SG, Scarascia- control. Transportation research board, National research coun-
Mugnozza G, Strom A and Hermanns RL (eds) Advanced research cil, Washington, D.C., pp 1133
workshop: landslides from massive rock slope failure. NATO Sci- Voight B (1981) Time scale for the first moments of the 1980 erup-
ence Series, IV Earth and Environmental Sciences, June 1621, tion. In: Lipman PW, Mullineaux DR (eds) The 1980 eruptions of
2002, Celano, Italy, 49, pp 285304 Mt St Helens, Washington. , US Geological Survey Professional
Melosh HJ (1997) Impact cratering. In: Shirley JH, Fairbridge RW Paper, 1250, pp 69134
(eds) Encyclopedia of planetary sciences. Chapman and Hall, Voight B, Glicken HX, Janda RJ, Douglas PM (1981) Catastrophic
London, pp 326335 rockslide-avalanche of May 18 1980. In: Lipman PW, Mullineaux
Prostka HJ (1978) Heart Mountain fault and Absaroka volcanism, DR (eds) The 1980 eruption of Mt. St. Helens, Washington. US
Wyoming and Montana, U.S.A. In: Rockslides and avalanches, Geological Survey Professional Paper, 1250, pp 347377
I.Voight B (ed): Developments in geotechnical engineering. 14A, Yalin MS (1971) Theory of hydraulic models. Macmillan, London
Elsevier, Amsterdam, 423437 Yarnold JC, Lombard JP (1989) A facies model for large rock ava-
Read SAL (1979) Lake Waikaremoana outlet engineering geologi- lanche deposits formed in dry climates. In: Colburn IP, Abbott
cal studies of factors related to leakage through the natural dam. PL, Minch J (eds) Conglomerates in basin analysis: a Sympo-
NZ Geological Survey report EG336 sium dedicated to A. O. Woodward. Pacific Section of Society of
Read SAL, Beetham RD, Riley PB (1991) Lake Waikaremoana bar- Economic Paleontologists and Mineralogists, 62:931
rier a large landslide dam in New Zealand. In: , Bell DH(ed) Zhang YQ, Lu Y, Hao H (2004) Analysis of fragment size and ejec-
Landslides. Balkema, Rotterdam, pp 14811487 tion velocity at high strain rate. Int J Mech Sci. 46:2734
Chapter 9

Residual Shear Strength of Tertiary Mudstone

and Influencing Factors

Binod Tiwari

Abstract. Extensive research done on the residual shear strength locations. Niigata Prefecture of Japan is one among the
of the Tertiary mudstone showed that mudstones are rich in ex- prefectures that have numerous dormant landslides trig-
pansive clay minerals. Those clay minerals are responsible for
gered on the mudstone formations. Shown in Fig. 9.1 is
making the mudstone highly weathering susceptible, which is the
main cause for excessive landslides in the mudstone formations. the county-wise distribution of landslides in Niigata Pre-
Because of high brittleness, residual shear strength is more im- fecture. Higashi Kubiki County of the Niigata Prefecture
portant in analyzing the landslides occurred in such formations. has the highest number of landslides (Tiwari et al. 2004).
Research results showed that average liquid limit, plasticity index, Approximately 91% of the county is covered with the
activity, and clay sized fractions in the mudstone from the Niigata
mudstone formation. More than 50% of the entire county
Prefecture of Japan are 72%, 36%, 1.4, and 26%, respectively. Domi-
nating clay mineral in mudstone is smectite with an average pro- was observed under dormant landslides. Those land-
portion of 15%. Dominating clay mineral oxide of mudstone was slides were observed in two types of mudstone forma-
aluminum oxide with an average value of 18%. Average residual tions Sugawa and Taruda formations. This paper deals
friction angle was observed to be 13. At the presence of saline with the properties of those mudstones pertinent to the
pore water the strength can increase up to 40%, depending on occurrence of landslides.
the nature of soil and site condition. If it is hard to get the soil
specimens in sufficient quantity from the mudstone area, residual
friction angles can be approximately estimated with index prop-
erties. However, if the mineralogical composition is known, residual
friction angle can be estimated with up to 90% accuracy, using
the diagram proposed by Tiwari and Marui (2005). This paper deals
with the chemical, mineralogical, and mechanical properties of the
Tertiary mudstone from the Niigata Prefecture of Japan, as well as
the methods to measure residual shear strength in conventional
soil testing devices.

Keywords. Landslides, mudstone, mineralogical composition, chemi-

cal composition, weathering index, residual shear strength, index

9.1 Background

Tertiary mudstones are over-consolidated sedimentary

rocks made of clays and silts of sedimentary origin.
Because of the tectonic activities and unequal settle-
ments, these rocks are highly fissured. Due to their
low strength and easily weathering nature, they are also
referred to as soft rocks. As they have low strength,
have low permeability, are expansive in nature, and
are very brittle, mudstones ate susceptible to shear
failure. We have evidenced a large number of pro-
gressive failure cases on the slope formed on mudstones.
In Japan, several places are formed with this type of
Tertiary mudstone. Natural hazards such as landslides, Fig. 9.1. County-wise Distribution of Landslides in the Niigata Pre-
and slope failures are frequently occurred on those fecture of Japan
136 Binod Tiwari

9.2 Properties of Tertiary Mudstone

Mudstones are made of expansive clay minerals that in-

clude smectite. Although they are stiff in nature, they
soften easily under the presence of water. Therefore, pro-
portions of expansive minerals play an important role in
the properties of the mudstone. To measure the proper-
ties of the Tertiary mudstone, numerous soil samples were
collected from a number of landslide sites in Higashi
Kubiki County (Fig. 9.2). Shown in Fig. 9.3 is the geologi-
cal map of the soil sampling area. Index properties of
those mudstone specimens are presented in Table 9.1. The
liquid limit, plasticity index, and clay fractions ranged
from 51 to 120%, 15 to 68%, and 3 to 45%, respectively,
and the average values were obtained to be 72.3%, 36.2%,
and 25.5% respectively.
Main dominating clay minerals in the Tertiary
mudstone are smectite and kaolinite. Nature of the
soil, liquid limit, activity, and shear strength of the soil
depend on the proportion of smectite in it. Shown in
Table 9.2 are the proportions of various minerals present
in the studied Tertiary mudstones. Those propor-
tions were measured by X-ray diffraction method. For
details of the X-ray diffraction, please refer Tiwari and
Marui (2005). Table 9.2 shows the range of minerals
in the soil specimens. According to the Table 9.2,
proportion of smectite, kaolinite, illite, and massive
minerals (quartz and feldspar) ranged from 7 to 24%,

Fig. 9.2. Location of the mudstone sampling area

Chapter 9 Residual Shear Strength of Tertiary Mudstone and Influencing Factors 137

Fig. 9.3. Geological map of the study area

2 to 15%, 0 to 3%, and 70 to 87%, respectively with

the average value of 15.2%, 5.8%, 0.7%, and 75.6%,
Mudstones are highly susceptible to physical weath-
ering. It is important to know the chemical compositions
of the soil specimens to predict the physical and chemi-
cal weathering behavior. Proportions of the major chemi-
cal oxides of several mudstone specimens are presented
in Table 9.3. According to the Table 9.3, Al2O3 is the ma-
jor chemical oxide pertinent to clay minerals, obtained
in the Tertiary mudstone. CIA (Chemical Index of Alter-
ation), CIW (Chemical Index of Weathering), PFA (Pla-
gioclase Index of Alteration), V (Vogts Residual Index),
R (Ruxton Ratio), WIP (Weathering Index of Parker), and
138 Binod Tiwari

STI (Silicate Titanium Index) of those specimens were

also calculated and are presented in Table 9.4. These pa-
rameters are referred to as the weathering indices. CIA
is calculated as CIW value of less than 50% shows that the rock is still
fresh, whereas CIW value of 100% shows the rock is in the
verge of complete weathering. Likewise, PIA is calculated as

CIA value of less than 50% shows that the rock is still
fresh, whereas CIA value of 100% shows the rock is in the PIA value of less than 50% shows that the rock is still
verge of complete weathering (Nesbitt and Young 1982). fresh, whereas PIA value of 100% shows the rock is in the
CIW is calculated as verge of complete weathering. V is calculated as
Chapter 9 Residual Shear Strength of Tertiary Mudstone and Influencing Factors 139

For fresh rock, V is less than 1, where as for the rock WIP value is more than 100% for the fresh rock,
close to complete weathering, V is infinite. Similarly, R is whereas WIP value is 0% for the completely weathered
calculated as rock. Likewise, ATI is calculated as

R values of fresh rock and completely weathered rock

are more than 10, and 0, respectively. WIP is calculated as
140 Binod Tiwari

STI value of more than 90% shows that the rock is still
fresh, whereas STI value of 0% shows the rock is in the
verge of complete weathering.
CIA, CIE, PFA, V, R, WIP, and STI in those specimen var-
ied from 67-79%, 74-90%, 71-80%, 1.8-3.4, 5.8-7.2, 36-51%,
and 81.7-83.6%, respectively. The average value of SiO2,
TiO2, Al2O3, Fe2O3, MnO, MgO, CaO, Na2O, K2O, and P2O5
are 68%, 0.7%, 18%, 6.6%, 0.04%, 2%, 0.7%, 1.5%, 2.5%,
and 0.13%, respectively. Likewise, average values of CIA,
CIW, PIA, V, R, WIP, and STI were 73%, 79.7%, 78.6%, 2.4,
9.1, 43.8%, and 82.9%, respectively.

9.3 Shear Strength Properties

As a result of brittleness in mudstone due to the over-

consolidation effect, there is a sharp drop in the strength
from peak to residual within a short displacement amount
(Skempton 1985; Tiwari and Marui 2005; Duncan and
Wright 2005; Mitchell and Soga 2005). Shown in Fig. 9.4
is a typical stress-strain diagram for the undisturbed and
remolded soil specimens from Okimi landslide. Due to a
number of fissures existing in the rock mass, fully soft-
ened or residual shear strength is mobilized during slid-
ing (Morgenstern 1978) in many cases. As a result of the
expansive nature of the mudstone and due to the exist-
ence of many dormant landslides in the mudstone areas,
it is important to evaluate the residual shear strength of
the soil than the peak strength. Table 9.5 shows the ranges
of peak, fully softened, and residual shear strength of a
number of mudstone specimens. Peak shear strength was
obtained by shearing the undisturbed specimen and the
fully softened shear strength was obtained by shearing the
remolded specimen both in the CD-triaxial tests. Residual
shear strength was measured in the ring shear device.

Fig. 9.4. Typical Stress-strain diagram for the tested mudstone speci-
142 Binod Tiwari

Fig. 9.8. Pre-cut mold and polished specimen used for the residual
shear strength measurement at triaxial device Fig. 9.10. Pre-cut soil specimen after shearing in triaxial compres-
Fig. 9.9.
Specially fabricated free platen
for the triaxial testing to mea-
sure residual shear strength
Chapter 9 Residual Shear Strength of Tertiary Mudstone and Influencing Factors 143

portion of smectite and specific surface area. Shown in

9.4 Factors Influencing the Residual
Figs. 9.11, 9.12, and 9.13 are those relations. The result
Shear Strength of Mudstone
shows that each type of dominant clay mineral gives
unique correlation curve, and the vertical position of the
As majority of the minerals in the soil from the mudstone curve depends qualitatively on the uniformity coefficient.
are smectite, kaolinite, illite, feldspar, and quartz, Tiwari Therefore, although residual shear strength could be es-
and Marui (2005) conducted a separate research to ob- timated approximately with the index properties, precise
serve the quantitative influence of mineralogical compo- estimation requires the exact proportion of constituent
sition on the residual shear strength. Tiwari and Marui minerals. Based on the soil test results from more than
(2005) compared the residual shear strength of the soil 120 natural and laboratory prepared soil specimen, Tiwari
specimens with index properties that include clay frac- and Marui (2005) proposed a triangular estimation chart,
tion, liquid limit, and plasticity index, as well as the pro- as shown in Fig. 9.14, which gives better estimation of
residual friction angle of the soil specimen based on the
mineralogical composition. Position of each soil speci-
men is plotted in the triangular estimation chart based
on the proportions of massive, low plasticity clay, and high
plasticity clay minerals and residual friction angle is esti-
mated by interpolating the corresponding contour lines.
Shown in Fig. 9.15 are seven different zones, where the
residual shear strength behavior changes with the domi-
nation of individual clay mineral. For the proportion of
smectite and kaolinite up to 8% each, residual friction
angle is controlled by the quartz and feldspar, which is
approximately 29. However, smectite proportion in the
Fig. 9.11. Relation between residual friction angle and liquid limit soil, if is more than 8%, controls the behavior of the soil
significantly. If the proportion of smectite in the soil speci-
men is more than 42%, residual shear strength of the soil
is approximately 4, irrespective of the proportion of the
other minerals. For other mineralogical composition, the
residual friction angle of a soil specimen depends on the
proportion of massive minerals (quartz, feldspar, calcite
etc), low plasticity clay minerals (kaolinite), medium plas-
ticity clay minerals (illite and chloride) and high plastic-
ity clay minerals (smectite). Approximate values of the
residual friction angle for different combinations can be
obtained by interpolating the contours in the chart pre-
sented in Fig. 9.14. Tiwari and Marui (2005) could esti-
mate the residual friction of mudstone specimens with
90% accuracy using the Fig. 9.14. Shown in Fig. 9.16 is
Fig. 9.12. Relationship between residual friction angle and plastic- the domination of kaolinite and smectite in each zone.
ity index Both smectite and kaolinite are dominant in zone A and
zone B. Kaolinite has less effect in zone C, D, and E.
Other than the mineralogical composition, pore water
chemistry plays an important role in the residual shear
strength of mudstone. When residual shear strength of
different minerals were measured with sea water as a pore
fluid, smectite showed a significant increase in residual
friction angle, i.e. 10 from the value of 4 (with distilled
water) (Tiwari et al. 2005). However, kaolinite did not
show much influence of sea water on residual shear
strength. This shows that pore water chemistry plays an
important role on the residual shear strength of mudstone.
Figure 9.17 shows the increase in residual friction angle
due to the presence of sea water with the proportion of
Fig. 9.13. Relationship between residual friction angle and clay fraction smectite. The increase in the residual friction angle with
144 Binod Tiwari

Fig. 9.16. Influence of smectite and kaolinite in residual shear strength

for various zone shown in Fig. 9.15

Fig. 9.14. A triangular diagram for the estimation of residual fric-

tion angle with mineralogical composition

Fig. 9.17. Increase in residual friction angle with proportion of

smectite while testing the specimen at saline water

ful while assessing the long-term stability of a mudstone

slope in the coastal area. Figure 9.18 shows the residual
friction angle of a number of mudstone specimens under
sea water, distilled water and after leaching.

9.5 Conclusion

Fig. 9.15. Different zones that have influence in residual friction angle Extensive research was conducted for about a decade to
of soil measure the residual shear strength of mudstones and
the influencing factors on the residual shear strength. The
saline water depends on the thickness of diffused double research showed that Tertiary mudstones are rich in ex-
layer (Tiwari et al. 2005; Mitchell and Soga 2005). Due to pansive minerals and are highly susceptible to strength
its formation process, mudstones near the coastal area softening in the presence of water. Due to its brittleness
are rich in NaCl concentration. A significant proportion characteristics, residual shear strength plays vital role in
of Na+ ion leaches out through the interaction with fresh the stability of the mudstone slope. Average values of the
water after being crushed out during sliding and residual liquid limit, plasticity index, activity, and clay fraction of
shear strength decreases continuously. This degrades the the mudstone from Niigata Prefecture of Japan were 72%,
stability of the slope. Therefore, we should be very care- 36%, 1.4, and 26%, respectively. Dominating clay mineral
Chapter 9 Residual Shear Strength of Tertiary Mudstone and Influencing Factors 145

Fig. 9.18.
Residual friction angle of the
soil specimen with distilled wa-
ter, saline water, and after leach-
ing the Na+

of the Tertiary mudstone was smectite with an average pro- Morgenstern N (1977) Slopes and excavations in heavily over con-
portion of 15%. Average residual friction angle of those solidated clays, In: Proceeding of 9th International Conference of
Soil Mechanics and Foundation Engineering, 2, pp 567581
mudstones was 13. The shear strength of the mudstone in
Nesbitt HW, Young GM (1982) Early proterozoic climates and plate
the field, where saline pore fluid was observed was about motions inferred from major element chemistry of lutites. Na-
40% higher than the mudstone after leaching Na+ ion. The ture 279:715717
research also shows that the residual shear strength can be Skempton AW (1985) Residual strength of clays in landslides, folded
estimated approximately with liquid limit, plasticity index, strata and the laboratory. Gotechnique 35(1):318
Tiwari B, Marui H (2003) Estimation of residual shear strength of
and specific surface area. However, better estimation can
Toyoura sand-Bentonite-Kaolin Mixture. J Jpn Landslide Soc
be performed using the diagram proposed by Tiwari and 40(2):124133
Marui (2005) if the mineralogical composition is known. Tiwari B, Marui H (2005) A new method for the correlation of re-
Quantity of soil specimen obtained from boring core is suf- sidual shear strength of the soil with mineralogical composition.
ficient to estimate the mineralogical composition. J Geotech Geoenviron 131(9):11391150
Tiwari B, Marui H, Aoyama K, Bhattarai P, Tuladhar G (2004) Prepa-
ration of geotechnical database of mountainous cities with
ArcGIS. In: Proceedings of ESRIs 24th Annual International Users
References Conference, San Diego, 1569, pp 130
Tiwari B, Brandon T, Marui H, Tuladhar G (2005a) Comparison of
Bishop AW, Green GE, Garga VK, Andresen A, Brown JD (1971) New residual shear strength from back analysis and ring shear tests
ring shear apparatus and its application to the measurement of on undisturbed and remolded specimens. J Geotech Geoenviron
residual strength. Gotechnique 21(4):273328 131(9):10711079
Chandler RJ (1966) The measurement of residual shear strength in Tiwari B, Tuladhar G, Marui H (2005b) Variation in residual shear
triaxial compression. Gotechnique 16(3):81186 strength of the soil with pore water chemistry. J Geotech Geo-
Duncan JM, Wright S (2005) Soil strength and slope stability. John environ 131(12):14451456
Wiley & Sons Inc., New York Tiwari B, Duncan JM, Brandon TL (2006) Evaluation of undrained
Mitchell JK, Soga K (2005) Fundamental of soil behavour. 3rd ed. John shear strength of clays in triaxial compression, submitted to Na-
Wiley & Sons, Inc., USA tional Science Foundation, Virginia Tech
Chapter 10

On Failure of Municipal Waste Landfill

Ikuo Towhata

Abstract. One of the serious problems in mega cities in developing ure of Payatas landfill in Metro Manila is famous as a rain-
countries is the management of municipal waste. Due to priority of eco- fall-induced failure of waste deposit (Merry et al. 2005). A
nomic development or insufficient attention to waste problems, many
similar accident, however, has occurred elsewhere.
mega cities simply dump waste in the field without provision for en-
vironmental and mechanical risks. An example of problems of this
type occurred in Bandung City of Indonesia in February, 2005, in which
a large waste landfill collapsed after rain fall and killed more than one 10.2 Failure of Waste Leuwigajah Landfill in
hundred people. Similar accident occurred earlier in the Philippines Bandung, Indonesia
as well. This text therefore makes a brief report on this event and
shows the need for more care for landfill operation and management.
Bandung, Indonesia, is a city which is located in West Java
Keywords. Landfill, slope failure, municipal waste, case study Island and has six-million population (Fig. 10.1). In this city,
the Leuwigajah Landfill was the biggest and most impor-
tant site for dumping waste. It was situated in a narrow val-
ley whose original gradient was 510% and its subsurface
10.1 Introduction conditions consisted of bedrock covered by 1-m thick clay.
The dumped landfill formed a waste deposit of 60 to 70 m
In the recent decade, many countries which used to be in thickness. It appears that all kinds of municipal waste of
called developing countries have made remarkable suc- the city were simply dumped without incineration or sepa-
cess in economic development. Accordingly, their popu- ration (Fig. 10.2). Hence, organic waste and plastics are
lation increased and the increased people started to live mixed in the fill. The landfill was constructed since 1992 by
in big cities. This situation resulted in emerging mega cit- simply dropping garbage from the top over an edge and
ies which have millions of population. The problem lying poor compaction was made by crawlers.
in those new mega cities is that economic development is The Leuwigajah landfill collapsed suddenly at 2 A.M.
given with the first priority and that there is not suffi- on February 21st, 2004, over an area of 0.13 km2 (Koelsch
cient provision for good environment. Hence, air and et al. 2005). The author visited the site in February, 2006,
water are polluted and poor traffic system worsens the which was one year after the accident. The present text is
congestion and air pollution. Environmental pollution written based on knowledge obtained during this visit.
made by industrial waste is another problem.
A problem which is similarly serious is the management
of municipal waste which is produced by peoples daily life
as well as offices and small industries. Since budget is not
sufficient for waste treatment, the municipal waste is often
dumped without taking care of accompanying risks. While
the risk of environmental pollution is frequently mentioned,
the present text concerns slope instability problem. The fail-

Fig. 10.1. Location of Bandung City Fig. 10.2. Appearance of remaining part of Leuwigajah landfill
148 Ikuo Towhata

The failure created a cliff at its top part (Fig. 10.3).

Being nearly vertical and having been stable for one year
after the failure, this cliff suggests that the waste material
has some shear strength, although the downslope part
failed. Figure 10.4 on the contrary demonstrates the view
towards the downstream direction. The failed waste
flowed over a long distance and covered an area of
300 900 m2. The apparent angle of friction as determined
from the vertical and horizontal travel distances was 6 de-
grees. The death toll due to this accident was 146. Local
people told the author that there was a heavy rainfall for
three days prior to the failure. Since the landfill was lo-
cated in valley topography, probably much ground water
came into the waste deposit and affected the stability.

Fig. 10.3. Head scarp of landfill failure

10.3 Significance of Landfill Failure

Similar to other landfills in South East Asia, many scav-

engers were living in the downslope area of the Leuwigajah
landfill. Their houses were too close to the landfill
(Fig. 10.5) and many of them were buried under the failed
waste mass. Scavengers pick up many kinds of waste
(Fig. 10.6) and sell for money. For example, 1 kg of plastics
are sold at 750 Rupia in the local currency; approximately
equal to 9 Japanese Yen or 6 US Cents. Although there was
no eyewitness of this midnight slope failure, the local
people mentioned that the landfill had been moving for
two days prior to the ultimate failure. This implies the
effects of rainfall which occurred in the meantime.
Figure 10.4 showed that the failed waste reached the
area of rice field. Although the quality of irrigation water
is affected by the waste, there is no effort to prevent the
Fig. 10.4. Widespread deposit of failed waste pollution problem.

Fig. 10.5.
Scavengers houses
Chapter 10 On Failure of Municipal Waste Landfill 149

Fig. 10.6. Scavenger who pick up waste Fig. 10.7. Houses which survived the failure of landfill

10.4 Proposals for Better Conditions 10.5 Conclusions

The ultimate solution for environmental and stability A brief report was made of a recent collapse of municipal
problems is obviously a construction of dam on the down- waste landfill in Indonesia. It was seen that safe but ex-
stream side of the dumping site. Such a measure is, how- pensive measures are not accepted due to economic rea-
ever, ruled out for economic reasons in many developing sons. Alternative safety measures are the location of
countries. Hence, alternative ideas are needed. houses in safer area, making gentle slope angle during
Firstly, a control of habitation is necessary. In the waste dumping, and installation of warning system and
Leuwigajah case, houses located in the valley slope drainage as well as waste treatment facility.
survived the accident (Fig. 10.7). It seems possible
to make a rough estimate of the run-out area of a pos-
sible landfill failure by using the aforementioned ap- Acknowledgments
parent friction angle of 6 degrees. Houses should be
located in areas which are not affected by the waste The authors visit to the failed landfill was made easy by
flow. Secondly, the dumping of waste should make assistance of Dr. F. Koelsch of Technical University of
a gentle slope gradient; formation of a steep slope Braunschweig, Germany, and Dr. Ilyas Suratman of
should be avoided. Warning system for slope instability Bandung Institute of Technology. The author expresses
is a useful idea, for which needed cost is not high. his sincere thanks to their collaboration.
Finally, good drainage system should be installed and
the leachate water should not affect the habitation
and agricultural areas. A water treatment facility seems References
to be a good measure to achieve this goal. Although
such a facility may be called expensive, it is a reasonable Koelsch F, Fricke K, Mahler C, Damanhuri E (2005) Stability of
landfills, the Bandung dumpsite disaster. In: Proceedings of the
target of international development aids. Mitigation
10th International Landfill Symposium
of many mega city problems in developing countries Merry SM, Kavazanjian E, Fritz WU (2005) Reconnaissance of the
should be considered as a useful and meaningful point July 10, 2000, Payatas landfill failure. Journal of Performance of
of international aids. Constructed Facilities, ASCE 19(2):100107
Chapter 11

Experimental Study with Ring Shear Apparatus on the May 2004

LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan

Fawu Wang* Kyoji Sassa

Abstract. In May 2004, a landslide occurred at the right flank of the eling for 72 km. The upper part of the central ridge sand-
Jinnosuke-dani landslide, and transformed into a debris flow after wiched by the Bettou-dani at the left side and the
fluidization. By analysis of the monitored video images of the de-
Jinnosuke-dani at the right side is an active landslide, ac-
bris flow, field investigation on the source area of the landslide, and
a series of simulation tests with a ring-shear apparatus on the ini- cording to data obtained by monitoring, and is called the
tiation of the rainfall-induced landslide and its traveling process, the Central Ridge Block of the Jinnosuke-dani landslide
initiation and traveling mechanisms of the debris flow traveling in (Fig. 11.3). The width and length of this block are 500 m
the valley were investigated. It is shown that concentrated ground- and 2 000 m respectively. Besides the Central Ridge
water flow was the main reason for the landslide initiation, and a rapid
Block, there are many other active landslide blocks in
decrease of the mobilized shear resistance even under naturally
drained condition caused the rapid landslide motion. During the this area. In Fig. 11.3, the blocks with arrow inside are
debris motion in the valley, high potential for grain-crushing of de- active landslide blocks. In recent decades, accompanying
posits in upstream and lower potential for the downstream deposits the deformation of the Central Ridge Block, local land-
controlled the traveling and depositing process of the debris flow. slides with different volumes occurred at both boundary
Different grain-crushing potential of the valley deposits played an valleys of the Central Ridge Block and caused damage,
important role in the debris flow traveling and depositing processes.
although many countermeasure works have been con-
Keywords. Landslide, case study, fluidization, groundwater, grain- ducted in this area for more than 50 years.
crushing, ring-shear test The Electronic Distance Measuring (EDM) method
and Global Positioning System (GPS) monitoring method
have been used at this site to monitor slope displacements.
Figure 11.4 shows the initial locations of the monitoring
11.1 Introduction points in the regulated Landslide Prevention Area and
the displacement vectors of the slope surface from 1994
Haku-san Mountain is located at the boundary between to 2001 (Wang and Sassa 2007). There are 9 survey points
Ishikawa Prefecture and Gifu Prefecture in Hokuriku dis- (A1 to A9) in the lower block of the landslide and 12 sur-
trict, Japan (Fig. 11.1). It is an active volcano with a sum- vey points (B1 to B12) in the upper block. Six points (C1
mit 2703 m in elevation, and the whole mountain is a na- to C6) were located outside of the Jinnosuke-dani land-
tional park. This park is famous for its beautiful scenery.
About 50 000 mountain climbers visit this mountain in
the period from 15 May to 15 October every year. Tedori
River, the largest river in Ishikawa Prefecture, originates
in this area. The Jinnosuke-dani landslide (Fig. 11.2) is a
giant landslide located on the southwestern slope of Haku-
san Mountain (dani means valley or torrent in Japa-
nese.). It was also the first landslide designated as a Land-
slide prevention area by the Japanese Landslide Preven-
tion Law in 1958. Landslides frequently occur in this area,
and commonly trigger debris flows that travel long dis-
tances and damage properties in the downstream valley
of the Tedori River. In the photograph of Fig. 11.2, the
areas not covered by vegetation are local slope failures.
For example, at the left side of the photograph, the main
scarp and sliding surface of the Bettou-dani failure, which
occurred in 1934, is visible. In that event, a debris flow
initiated by a landslide reached the Japan Sea after trav- Fig. 11.1. Location map of the Jinnosuke-dani landslide
152 Fawu Wang Kyoji Sassa

Fig. 11.2.
Aerial photograph of the Land-
slide Prevention Area of the Jin-
nosuke-dani landslide (photo
courtesy of Kanazawa Office of
Rivers and National Highways,

Fig. 11.3.
Active landslide blocks in the
Haku-san mountain area around
the Central Ridge Block of the
Jinnosuke-dani landslide (based
on Kanazawa Office of Rivers and
National Highways, MLIT 2004a)
Chapter 11 Experimental Study with Ring Shear Apparatus on the May 2004 LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan 153

Fig. 11.4.
Cumulative displacements of the
landslide from 1994 to 2001. All
the points of A, B, and C are slope
surface displacement monitor-
ing points

Fig. 11.5.
Photo of debris retention dams
constructed in the Jinnosuke-
dani valley

slide (It is used to monitor the activity of the other land- direction of movement corresponds well with the dip
slide blocks in the regulated landslide prevention area). direction of the Tedori Formation.
The monitored results show that the upper block dis- Figure 11.5 shows the debris retention dams con-
placed quite actively; the cumulative displacements of structed in the Jinnosuke-dani valley. In May 2004, a land-
survey points B5 and B11 exceeded 1 100 mm in the slide occurred at the upper part of the Bettou-dani from
7 years. The main features of the monitoring results in the Central Ridge Block. This landslide was transformed
the upper block are: (1) the points at the central part al- into a debris flow that traveled more than 2 km after it
most moved along the downslope direction; (2) the points slid into the Bettou-dani. A suspension bridge was com-
near valleys had a component to the valley side, besides pletely destroyed, and a local road with a simple bridge
along the downslope direction. However, the lower block utilized for debris-retention dam construction at the
has been relatively stable because the monitored displace- middle of Bettou-dani was heavily damaged. Fortunately,
ment at A1 to A9 at a rate of 315 mm yr1. While, C1, C2, nobody was injured because there were not many moun-
C3 and C4 which located on different landslide blocks tain climbers passing through the valley when the land-
also indicated the motion of the corresponding blocks. slidedebris flow occurred. However, because landslides
The boundary between the upper block and the lower with similar behavior frequently occur in this area, the
block is not clear at the slope surface. So, the landslide is risk for further landslide and debris flow activity still ex-
divided to blocks just according to the surface displace- ists. As a national park, it should be absolutely safe for
ment. The average movement direction of the upper the tourists. Even if some large landslides cannot be com-
block of the Jinnosuke-dani landslide is S 36 W. This pletely stabilized, understanding their potential risk, es-
154 Fawu Wang Kyoji Sassa

pecially their motion behavior is also very important for

disaster mitigation. Looking at the large parks in the
mountainous areas in the world, it is easy to find a com-
mon feature, that is, steep topography with nice views of-
ten has high risk for landslide. This paper attempts to
clarify the initiation and traveling mechanisms of the
landslidedebris flow, aiming to supply insight for future
landslide disaster prevention in similar area.

11.2 General Conditions of the Jinnosuke-dani

Landslide on Haku-san Mountain

The Haku-san Mountain area is characterized by heavy

precipitation and the Tedori River is characterized by its
Fig. 11.6. Hourly rainfall before the landslidedebris flow occurred
(The rainfall gauge was located at the central ridge block of
steep gradient (Wang et al. 2004; Okuno et al. 2004). In
Jinnosuke-dani landslide, and was measured by Kanazawa Office of winter, due to the strong influence of monsoons from Si-
Rivers and National Highways, MLIT) beria, the accumulative snowfall may exceed 12 m in the
Haku-san Mountain area. In other seasons, half of the days
are rainy. For this reason, local annual average precipita-
tion is 3295 mm, about two times the national average of
1700 mm for Japan. In this area, snowmelt generally be-
gins in the middle of March and finishes at the end of
May. In May 2004 when the landslide occurred, snow
melting was on-going, and the accumulative rainfall for
three days before the landslide occurred was 216 mm
(Fig. 11.6). It is also reasonable to consider that the effec-
tive water that infiltrated into the slope should be more
than this value. Because the movement was recorded by a
video camera of the Kanazawa Office of Rivers and Na-
tional Highways, Ministry of Land Infrastructure and
Transport, Japan (KORNH- MLIT) (2004b), the actual fail-
ure time was also exactly recorded, and the video image
is very important for the study on the motion mechanism
of the landslide and debris flow.
The basal bedrock in the Haku-san Mountain area is
the lower Paleozoic Hida gneiss. As a part of the 1 : 50 000
geological map of Haku-san Mountain area, a geological
map of the Jinnosuke-dani landslide and nearby area was
completed by Kaseno (2001). From the Jurassic to Early
Cretaceous periods, the Haku-san Mountain area was a
lake near the sea. The series of lacustrine sediments de-
posited in that period is called the Tedori Formation de-
posits. The deposits are sedimentary strata consisting of
shale, sandstone, and conglomerate layers that have un-
dergone hydrothermal alteration during the mountain-
building process of Haku-san Mountain. General descrip-
tions of the geology can be found in Kaseno (1993).
Figure 11.7a is a DEM model for a large area around the
Haku-san mountain area. The model is built on the el-
evation data with contour difference of 50 m. The image
Fig. 11.7. Topography and geological condition of the Haku-san and of lava deposit distribution around the summit of Haku-
nearby area. a DEM model; b geological map (after Kaseno 1993);
c longitudinal section of ZZ ' in (b) (after Kaseno 1993). T2: Sand-
san (vent of the Haku-san volcano) is visible. Figure 11.7b
stone and shale in Kuwashima group (also belong to the Tedori For- is the geological map of the corresponding area to
mation); Gr?: possible granite Fig. 11.7a. The Nohi Rhyolite (NR) of the Cretaceous pe-
Chapter 11 Experimental Study with Ring Shear Apparatus on the May 2004 LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan 155

riod is distributed at the upper right corner in the figure the dotted line area, the disconnection of H2 old volcanic
and the alternating layers of sandstone and shale of the deposit can be observed. The reason for this phenom-
Tedori Formation (T3) are distributed below and to the enon maybe caused by erosion and landslide.
left. Both units form the bedrock of the Haku-san Moun- Figure 11.8 shows the distribution of strata in the area
tain area. Volcanic lava deposits consisting of calc-alka- adjacent to the Jinnosuke-dani landslide. Alternating
line orthopyroxene and hornblende andesite, which layers of sandstone and shale of the Tedori Formation
erupted 100 000 and 10 000 years ago (H2 and H3, respec- are distributed at the left side of the figure, and the Cre-
tively), overlie the strata of the Tedori Formation and the taceous Nohi Rhyolite are distributed at the right side.
Nohi Rhyolite. Figure 11.7c shows an estimated longitu- Both form the bedrock of this area. Volcanic deposits,
dinal section of the ZZ' section (shown in Fig. 11.7b). which erupted 100 000 years ago and 10 000 years ago,
The deposition subsequence of the strata and volcanic overlie the strata of the Tedori Formation and the Nohi
deposit is clear. The Jinnosuke-dani landslide area is in- Rhyolites.
dicated in the dotted line box. At the northwest corner of

11.3 The May 2004 LandslideDebris Flow

As mentioned earlier, the landslide occurred on 17 May

2004 after continuous intense rainfall for two days. The
elevation of the source area of the landslide was about
1900 m, and the elevation of the toe part of the deposit of
the debris flow caused by the landslide was about 1200 m.
Figure 11.9 shows two aerial photographs taken before
the event (Fig. 11.9a: in the fall of 2003), and after the event
(Fig. 11.9b: on 24 May 2004, 7 days after the landslide-
Fig. 11.8. Geological map of the area adjacent to the Jinnosuke-dani debris flow), and Fig. 11.9c, the trace of the debris flow
landslide (modified from Kaseno 2003) with elevations at some key points. As shown in Fig. 11.9a,

Fig. 11.9.
The May 2004 landslidedebris
flow, which occurred in the
Bettou-dani from the Central
Ridge Block of the Jinnosuke-
dani landslide. a Aerial photo-
graph taken before the slope
failure (in the fall of 2003);
b aerial photograph after the
landslide (taken on 24 May 2004);
c trace of the debris flow (photos
courtesy of the Kanazawa Office
of Rivers and National High-
ways, MLIT)
156 Fawu Wang Kyoji Sassa

the source area is a steep cliff and there was not any veg- reached a maximum of 20 m s1. As shown in Fig. 11.9c,
etation on the lower segment of the landslide; however, at the relative height difference between the source area and
the upper part, the slope is relatively gentle and is cov- the toe of the deposits of the debris flow (near No. 10 de-
ered by vegetation. For mountain climbers, after leaving bris retention dam) was 700 m, and the horizontal travel-
Bettou Deai, which has facilities such as parking areas, ing distance was about 2000 m. Based on these data, the
rest rooms, simple restaurants, and a bus stop, most of apparent friction angle a (defined as tan a = H/L, where
the climbers have to cross the suspension bridge and ac- H is the difference of elevation between head and toe of a
cess the Central Ridge Block of the Jinnosuke-dani land- landslide, and L is the horizontal distance from head to
slide to get to the summit of Haku-san Mountain. At the toe) of the debris flow is estimated to be 19.3 degrees.
middle of the Bettou-dani, an access road for construc- Figure 11.10 shows the situation when the suspension
tion of debris-retention dams crosses the valley and en- bridge was completely destroyed. Large boulders 34 m in
ters the Central Ridge Block. As shown in Fig. 11.9b, both diameter were transported and deposited near the bridge
the roads and the bridge were badly damaged when the site. Some thin debris were deposited on the top of the left
debris flow hit them. The entire flowing process of the pillar of the bridge, about 10 meters above the valley bot-
debris flow was recorded by a video camera (see Fig. 11.9c tom. This shows that even near the terminus of the de-
for the photo of the video camera set at the site), which bris flow, the sliding potential was high and powerful.
was set at an elevation of about 1860 m for the purpose to Figure 11.11 shows a series of continuous images taken
monitor the debris flow in the Bettou-dani valley by the from the monitoring video of KORNH-MLIT. The loca-
KORNH-MLIT (2004c). It Japan, under the leadership of tion of the video camera was about 250 m downstream
the Ministry of Landslide, Infrastructure and Transport, from the source area of the landslide. The time in sec-
monitoring video cameras are set at different parts of onds is shown at the top of each image. Figure 11.11a
important rivers, especially at the parts which have high shows the situation just before the debris flow arrived.
risk for landslide and debris flow. The video camera moni- White dotted lines in these images are the boundary of
toring was conducted with a remote control system, and the sliding mass in the Bettou-dani valley. The white color
the video image can be reviewed in real time on the in the images is snow. The debris flow passed through
internet. Because Bettou-dani is a valley with high po- the video from 16:32:37 to 16:33:16 oclock; thus, the en-
tential for landslide and debris flow, it has been under tire process continued for only 40 seconds in front of the
control and monitored for 24 hours a day. Through ana- video camera. By analysis of the video images shown in
lyzing the recorded video images of this event, it is esti- Fig. 11.11, the debris flow can be divided into four sepa-
mated that the velocity of the debris flow may have rate waves. The first wave was from (b) to (g), which con-

Fig. 11.10. The suspension bridge that was completely destroyed by the May 2004 debris flow in the Bettou-dani (photo courtesy of the
Kanazawa Office of Rivers and National Highways, MLIT)
Chapter 11 Experimental Study with Ring Shear Apparatus on the May 2004 LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan 157

Fig. 11.11. Continuous images of the May 2004 debris flow in the Bettou-dani (video courtesy of the Kanazawa Office of Rivers and National
Highways, MLIT). Sliding mass is shown in white dotted lines. White color in the white dotted lines is snow

tinued for 7 seconds. As a frontier of the sliding mass, passed the video very quickly. The third wave was from
there was fog going along with the sliding mass, indicat- (i) to (s), which continued for 11 seconds. During this long
ing a high traveling speed of the sliding mass. The sec- period, the sliding mass passing before the video was very
ond wave was from (g) to (i), which continued for 3 sec- high and wide, indicating a large volume. As shown in
onds. In this short period, a relative small sliding mass Fig. 11.9s, an interrupt can be seen in the middle of the
158 Fawu Wang Kyoji Sassa

sliding mass. Actually, the flow of the sliding mass should exited at W1, W2, and W3 at relatively high positions. In
keep continuous at the lower position of the valley (From the other hand, at the lower part of L3, there were three
the camera location, the bottom of the valley cannot be groundwater exits W4, W5 and W6. These groundwater
observed). From 16:32:58, the dimension of the flow be- exits were located in relative low position comparing with
came smaller, but showing a continuous flow. For this rea- W1, W2 and W3. It looks like that the groundwater level
son, the fourth wave was defined from (r) to the end of was different at the left and right parts, and the exits indi-
the motion (at 16:33:16), which continued for 18 seconds. cated the different groundwater veins. It is estimated that
The images after (x) were not presented here because the the phenomenon of the sliding mass in L3 not moving so
direction of the video camera was changed to observe the far is due to that the sliding mass at L3 was not fully satu-
situation of the upstream part. For all of the images in- rated by the groundwater flow when the landslide oc-
cluding the sliding mass, it is obvious that all of the de- curred. This fact ensured that the debris, especially in the
bris included snow, and muddy fog can be recognized in potential sliding zone, was fully saturated and that high
Fig. 11.11d,e,r,s and 11.11t, indicating rapid motion dur- water pressure was supplied to the back of the debris to
ing downstream traveling. make the slope unstable. The groundwater exiting at a
Figure 11.12 shows the situation at the source area of high position at the head of the debris was a major trig-
the landslide on 11 September 2005, more than one year gering factor for the landslide which fluidized into a de-
after the landslide event. A man in the enlarged box can bris flow. It needs to be mentioned that, when the field
be seen as a scale. According to the report of investiga- investigation was conducted in September 2005, more
tion conducted soon after the event by KORNH-MLIT than one year had passed since the landslide occurrence.
(2004b), the average slope angle was about 28 degrees, and Some situation may be changed by rainfall erosion and
the average thickness of the sliding mass was estimated other factors. However, the basic condition should be the
as 30 m. In this figure, L1, L2, and L3 show the rear bound- same because there was no large change in natural condi-
aries of three different sliding blocks, which moved for a tion and no any engineering activity on this site. The dif-
limited distance from the main scarp; however, most of ference is that, landslide occurred in the snow melt sea-
these blocks did not move so far, but just rested on the son combining with heavy rainfall, and September is not
slope. At the middle block, between L1 and L3, most of rainy season. Considering this difference, it can only be
the debris material slid out of the source area, entered estimated that the flow at the groundwater exits should
into the Bettou-dani, and joined the debris flow. Also, at be stronger than that at the investigation.
the lower part of L2, most of the debris slid out into the To investigate the initiation and traveling mechanism
valley. A common phenomenon at the source areas of of the landslide-debris flow, soil samples were obtained
these sliding blocks is that concentrated groundwater flow at the source area and along the travel path in the valley

Fig. 11.12. Source area of the May 2004 landslidedebris flow in the Bettou-dani. Blue points represent the groundwater exits. Solid lines are
the rear boundaries of the sliding blocks (photo taken by F.W. Wang on 11 September 2005). Points W1W6 shows groundwater exits,
and S is the sampling point. A man in the enlarged box can be seen as a scale
Chapter 11 Experimental Study with Ring Shear Apparatus on the May 2004 LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan 159

of the Bettou-dani. Soil sample Beto-1 was taken from

point S (Fig. 11.12) at the source area. This sample was
subjected to ring-shear test, simulating a rainfall-induced
landslide, to investigate the initiation mechanism of the
landslide. Soil samples Beto-2, Beto-3, and Beto-4 were
taken from the surface of the valley deposits along the
traveling path of the debris flow, and were subjected to
ring-shear tests that simulated dynamic loading of the
landslide mass on the valley deposits and the dynamic
loading of the debris flow on the valley deposits to inves-
tigate the traveling mechanism of the debris flow travel-
ing down the valley.
In the soil sampling, coarse grains larger than 20 mm
were excluded because it is impossible to use them in the
laboratory tests. Figure 11.15 shows the grain-size distri-
bution of the four soil samples. Sample Beto-1 is the fin-
est sample among them, and all of the samples show a
similar gradation. For sample Beto-2, 3, 4, Beto-3 has the
least amount of fines. The grain-size distribution of the
samples may indicate the potential of the water transport
in the valley, i.e., upstream of the Beto-3 sample site, the
fine part dominates because of the supply of the weath-
ered material; while the downstream part is rich in fines,
because of their transport by water. The average grain-
sizes of the soil samples Beto-1, Beto-2, Beto-3 and Beto-
4 were 2.7 mm, 6.0 mm, 4.8 mm and 6.0 mm, while the
uniformity coefficient was 20.9, 25.7, 7.6 and 34.7, respec- Fig. 11.13. Design concept of ring shear apparatus (from Sassa et al.
tively. The specific gravity Gs of the samples was 2.71. 2004a)

11.4 Ring Shear Tests on the Initiation and Traveling them. This system is quite durable in regard to shearing
Mechanisms of the LandslideDebris Flow and is sensitive to pore-pressure monitoring, although the
monitoring point is not at the center of the shear zone
Figure 11.13 shows the design concept of the ring shear (Sassa et al. 2004a).
apparatus, by which the initial normal stress, initial shear The diameters of the outer ring and inner ring are
stress and the initial pore pressure acting on a soil ele- 180 mm and 120 mm, respectively. The sample, after
ment at sliding zone can be simulated. The major pur- placement in the shear box, had a donut shape with a
pose and the design concept is to geotechnically simulate width of 30 mm. To avoid possible grain-size effects on
the formation of the shear zone and the post-failure mo- the shearing behavior, only grains with diameter smaller
bility of high-speed landslides and observes the conse- than 4.75 mm were included in the tested samples.
quence of mobilized shear resistance, as well as the post-
failure shear displacement and generated pore-water pres-
sure (Sassa et al. 2004a). In this study, ring-shear tests were 11.4.1 Grain-crushing Susceptibility of the Valley
conducted using ring-shear apparatus DPRI-5, which was Deposits in Different Parts of the Bettou-dani
developed by Sassa in 1996 (Sassa et al. 2003).
One of the most important features in the development It is believed that the difference in grain-crushing sus-
of the ring shear apparatus is the development of an ef- ceptibility should cause the difference in the traveling
fective and durable pore-pressure monitoring system. To process of the landslidedebris flow. When the soil is
have a large inlet section and provide an average pore- easy to be crushed, excess pore pressure will be easily
pressure value throughout the soil sample, pore-pressure generated during the shearing process under rapid mo-
transducers are connected to a gutter (4 4 mm) extend- tion (means undrained condition), and in turn, the shear
ing along the entire circumference of the inner wall of resistance will decrease and result in high mobility of
the outer ring in the upper box, as shown in Fig. 11.14. landslides. To check the grain-crushing susceptibility of
The gutter is located 2 mm above the shear surface and is the valley deposits, dry ring-shear tests were conducted
covered by two metal filters, with a filter cloth between on the samples. The test conditions were: consolidate
160 Fawu Wang Kyoji Sassa

Fig. 11.14.
A half section of the shear box
and the close-up diagram of the
edges (from Sassa et al. 2004a)

Fig. 11.15. Grain-size distribution of soil samples taken from the Fig. 11.16. Sample-height change with shear displacement during con-
source area of the 2004 landslide and from the traveling path of the stant-shear-speed dry ring-shear tests on samples Beto-2, Beto-3, Beto-
debris flow 4, and Toyoura silica sand. Normal stress = 300 kPa, Shear velocity
= 10.0 mm s1. The void ratios for the four samples after consolida-
the sample at 300 kPa normal stress which corresponds tion (before shearing) are 0.687, 0.760, 0.753 and 0.901, respectively
to the actual stress level of the landslide, and shear it
under constant speed of 10.0 mm s1 until the shear dis- ceptibility generally has large sample-height change (con-
placement reaches 6.4 m. For comparison, Toyoura traction) during dry shear. The sample-height changes
silica sand which is known as a standard sandy soil that that occurred in the samples taken from Bettou-dani were
is difficult to crush, was also sheared under the same test quite a bit larger than that of Toyoura silica sand. Among
conditions. the samples taken from Bettou-dani, Beto-4 has the small-
Figure 11.16 shows sample-height change during the est sample-height change during shearing, showing a rela-
dry ring-shear tests. Soil with high grain-crushing sus- tively lower grain-crushing susceptibility.
Chapter 11 Experimental Study with Ring Shear Apparatus on the May 2004 LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan 161

0 = t hcos2 and 0 = t hsin cos, where t is the united

weight of the sliding mass, a average value of 18 kN m3
was assumed in the test. Assuming the groundwater table
near the sliding surface, the initial effective normal
stress (0) and shear stress (0) acting on the sliding sur-
face were 420 kPa and 224 kPa, respectively. Then, pore-
water pressure acting on the element increased as the re-
sult of rainfall and snowmelt. The test was conducted
under the following procedure:

1. saturate the soil sample to a high degree of saturation

with carbon dioxide and de-aired water; it was con-
firmed that the BD value (BD is a pore pressure param-
eter, related to the degree of saturation in the direct-
shear state. It was proposed by Sassa (1988)) reached
0.96, showing a high degree of saturation;
2. consolidate the sample under normal stress of 420 kPa;
3. apply the initial shear stress of 224 kPa gradually at
41.7 Pa s1 to avoid pore-water pressure generation;
4. increase the pore-water pressure gradually at the rate
of 0.5 Pa s1 through the upper drainage line directly
connecting to the upper surface of the sample until
failure occurs;
5. measure the residual friction angle of the soil with
constant shear speed, while increasing the normal
stress gradually from a low stress level (about 90 kPa)
to a high stress level (about 400 kPa).

Figure 11.18ac presents the test results. Fig. 11.18a

shows the time-series data for the whole test series. From
the beginning to nearly 200 s, the normal stress and shear
Fig. 11.17. Grain crushing occurred in the dry constant-shear-speed resistance were kept constant, while the pore pressure was
ring-shear tests on samples Beto-2, Beto-3, Beto-4, and Toyoura
silica sand. a Grain-size distribution of the tested samples before
increased gradually. From 200 s to 330 s, small displace-
and after shearing; b Marsals grain crushing susceptibility Bp (de- ment occurred, and the shear resistance mobilized a little
fined by Marsal 1967) bit higher, although the shear stress was kept constant.
The void ratio at 250 s was 0.350. After 330 s, rapid failure
Figure 11.17a,b shows the grain-size distribution of the occurred, which can be confirmed by the acceleration of
tested samples before and after shearing, and the grain- the shear displacement. Corresponding to the rapidly in-
crushing percentage Bp of all samples (Marsal 1967), respec- creasing shear displacement, the shear resistance de-
tively. Bp is the summation of the difference of grain-size creased rapidly to about 110 kPa. At that point, the appar-
distribution at each sieve size of the sample before and af- ent friction angle became 15.1 degrees, which is shown by
ter dry shear (taken from shear zone), and it indicates the the total stress path and effective stress path in Fig. 11.18b.
grain-crushing susceptibility of the soil. It is obvious that The residual friction angle of the soil was measured un-
grain-crushing susceptibility becomes lower from the up- der complete drained condition as 33.7 degrees, which is
stream part to the downstream part of the Bettou-dani. shown in Fig. 11.18c, when the above test was finished.
The procedure was to increase the normal stress gradu-
ally (5.6 Pa s1) from about 95 kPa to 405 kPa when keep-
11.4.2 Ring-shear Tests on Soil Samples ing a constant shear velocity. The shear velocity was set
Taken from the Source Area as slow as 0.02 mm s1 to avoid excess pore pressure gen-
eration in the shear zone. From the above test results, it is
The purpose of this test is to simulate the initiation of the easy to calculate that the slope can remain stable at its
landslide by water pressure. The initial slope condition initial slope angle of 28 degrees, if there is no increase in
was simplified as being 30 m in thickness (h) and 28 de- pore pressure at the sliding surface. In addition, from
grees in slope angle (). The landslide was triggered in Fig. 11.18b, it can be seen that the peak friction angle of
the ideal slope by rainfall and snowmelt water. Through the soil at initial failure is much higher than 33.7 degrees.
162 Fawu Wang Kyoji Sassa

pressure value from shear zone and upper drainage line

is difference, the high value will be recoded by the trans-
ducer. It is obvious that the high portion of the pore pres-
sure (measured) was from shear zone. From this differ-
ence, it is estimated that high excess pore pressure was
generated in the rapid motion process after the sample
failure. Related to the above concept, the pseudo effective
stress path (EPT) in Fig. 11.18b is not the actual effective
stress state in the shear zone, because the measured pore
pressure is not directly from the shear zone. It was also
affected by the supplied pore pressure. When pore pres-
sure at shear zone is correctly measured, the effective
stress path should move along peak or residual failure line.
In the later part of this test, only the shear resistance
should be replied on for data explanation.
Through the above test aiming to simulate the failure
process of a natural slope when the pore pressure acting
on the potential sliding surface was increased by the
groundwater table rising caused by heavy rainfall and
snowmelt, it is found that high excess pore pressure could
have resulted, and in turn, the shear resistance of the soil
at sliding surface could drop down rapidly. The resultant
rapid drop-down of the shear resistance should cause ac-
celeration of the landslide motion when it moved down
to the valley, and resulted in a high velocity when it rushed
into the valley.

11.5 Ring-shear Tests on Soil Samples Taken from

the Landslide Travel Path in the Bettou-dani

The above test simulated the initiation of rapid landslide

from natural slope. What will happen when the failed slid-
ing mass rushed into the Bettou-dani valley, where thick
torrent deposits were distributed? To simulate the land-
slide motion in the Bettou-dani, three other samples
(Beto-2, Beto-3, Beto-4) from different portion of the
Bettou-dani were sampled and used in ring-shear tests to
Fig. 11.18. Simulation test results of landslide initiation triggered by show the fluidization process of the landslide (see Fig. 11.9b).
rainfall under naturally drained condition. BD = 0.96, pore water pres-
sure increasing rate = 0.5 kPa s1. a Time-series data; b total stress
Beto-2 was taken near the uppermost debris-retention dam
path (TSP) and pseudo effective stress path (ESP); c residual fric- in the Bettou-dani; Beto-3 was taken near the destroyed
tion angle of the tested soil sample Beto-1 from the source area bridge; and Beto-4 was taken below the suspension bridge
and near the toe of deposit of the debris flow.
In Fig. 11.18, there are some concepts that need to be Figure 11.17 is a model proposed by Sassa et al. (1997)
clarified. At first, naturally drained condition means that to simulate the undrained-loading behavior of valley de-
in the test procedure drainage is not prevented and ex- posits by a rapidly sliding mass. The slide mass moved
cess pore-water pressure can generate depending on ma- down the slope (I), and applied load onto the torrent de-
terial behavior and loading rate, and it was referred to as posits at the foot of the slope (II). Because a surface wa-
naturally drained conditions by Sassa et al. (2004b). Un- ter stream or subsurface flow existed and some of the
der the naturally drained condition, it is easy to find out deposits were saturated, the torrent deposit was sheared
in Fig. 11.18a, that there is a difference between the pore by undrained loading and transported downstream to-
pressure (applied) and pore pressure (measured). As gether with the sliding mass (III). Here, it is assumed that
shown in Fig. 11.14, the pore pressure transducer is con- a column with unit length of its sliding surface, which is a
nected with the shear zone, also with the upper drainage part of the torrent deposit. In the position (I) of the slid-
line, which pore pressure was supplied. When the pore ing mass, the weight of the column (W0) was in effect.
Chapter 11 Experimental Study with Ring Shear Apparatus on the May 2004 LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan 163

When the sliding mass rode on to the torrent deposit (II) In Fig. 11.19, the valley deposit has a thickness of h0,
with a certain velocity, it provided dynamic loading of an initial slope angle of the valley , and groundwater
the column. It is assumed that the applied stress on the thickness on the sliding surface hw. The undrained load-
torrent deposits was as the sum of the static stress, W, (load ing from a rapidly moving displaced landslide has a thick-
due to the weight of the sliding mass) and the dynamic ness of h, an intrusion angle of , and a dynamic (im-
(impact) stress, Fd, working in the direction of motion of pact) coefficient of Kd. Based on Sassa et al. (1997) and
the sliding mass. Sassa et al. (2004a), it is reasonable for Kd to take a value
At the Bettou-dani, the sliding mass moved down the of unity. Then, the increment of normal stress and shear
slope (I), and applied a load to the valley deposits at the stress from the rapidly sliding mass to the deposits can
foot of the slope (II). Because a surface-water stream or be determined. The initial conditions for the three sam-
subsurface flow existed and some of the deposits were pling points, which were employed in the ring-shear tests,
saturated, the valley deposit was sheared by undrained are summarized in Table 11.1. From the upstream to
loading and transported downstream together with the downstream, the slope angle was changed from 18 degrees
sliding mass (III) (Sassa et al. 2004a). The above test to 5 degrees near the No. 10 debris retention dam. The
shown in Fig. 11.18 corresponds to the landslide that oc- initial thickness of the torrent deposits at the valley was
curred at slope (I). assumed to be the same value of 5 m, and the thickness of
To simulate the succeeding process, sample Beto-2 was groundwater was assumed to be 3 m (groundwater table
used to simulate the situation at slope (II), while samples was 2 m under below the torrent surface). The intrusion
Beto-3 and Beto-4 were used to simulate the behavior at angle was assumed as 10 degrees for Beto-2 sample near
slope (III), and the local slope angles of the valley at the the source of the landslide from natural slope. It is the
sampling points (Beto-3 and Beto-4) were considered. difference between the slope angle of the initiated land-
slide and the slope angle of the valley bed. When sliding
mass moved along the Bettou-dani valley, the intrusion
angle became to zero for Beto-3 and Beto-4. Because of
the effect of the debris retention dam, the thickness of
the sliding mass became thinner and thinner. Consider-
ing this phenomenon, the thickness of the sliding mass
was assumed to be 30 m, 20 m and 5 m from upstream to
Figure 11.20 shows the results of the simulation test
on slope (II) using sample Beto-2. Fig. 11.20a shows the
input stress signals of normal stress and shear stress be-
fore (05 s), during (515 s), and after the dynamic im-
pact process (1530 s). The signal was loaded on the
sample under the undrained condition to simulate the
rapid loading of sliding mass on the torrent deposits,
which had initial normal stress and initial shear stress
acting on it.
As can be seen in the time-series data (Fig. 11.20b),
pore-water pressure was generated at the same rate as the
applied normal stress, and failure occurred as soon as the
loading was applied. Shear resistance reached its peak
strength at 7 s, and arrived at its steady-state strength at
15 s. As shown by the stress paths (Fig. 11.20c), the ap-
parent friction angle is only 2.6 degrees, showing a high

Fig. 11.19. Model for undrained loading of saturated deposits by a

displaced sliding mass (Sassa et al. 1997). a Illustration of the model;
b stress path of the torrent deposit during loading. : Angle of thrust
between the slope and the torrent bed; Fd: dynamic stress; kd: dy-
namic coefficient (Fd/W)
164 Fawu Wang Kyoji Sassa

Fig. 11.20. Simulation test results on sample Beto-2 when sliding Fig. 11.21. Simulation test results on sample Beto-3 when debris flow
mass rushed into the valley and loaded on the torrent deposits. traveled along the valley in the middle part of the Bettou-dani val-
a Applied-stress signals (normal-stress and shear-stress incre- ley. a Applied-stress signals (normal stress and shear stress incre-
ments); b time-series data; c effective stress path (ESP) and total ments); b time-series data; c effective stress path (ESP) and total stress
stress path (TSP). BD = 0.96 path (TSP). BD = 0.99

mobility of the valley deposit after the dynamic loading. value as the slope angle of the Bettou-dani at this part.
All the series test results show that the valley deposits For this reason, the shear displacement generated in this
fluidized after the dynamic loading of the rapidly mov- test was only 28 mm when the loading was completed.
ing displaced sliding mass in an undrained condition. As a summary of the above dynamic tests, the shear
Figure 11.21 shows the results of the simulation test resistance at steady state under the undrained condition,
on sample Beto-3 at slope (III). The intrusion angle was the minimum apparent friction angle to show the mobil-
assumed to be zero because the displaced sliding mass ity of the torrent deposit loaded by rapid moved sliding
came from the upper part of the same valley with the same mass, and the residual friction angle of the three soil
slope angle. Fluidization also occurred, and the apparent samples taken from the Bettou-dani are presented in
friction angle mobilized at the steady state came to 4.4 de- Table 11.2. Because all of the possible grain-crushing
grees, slightly higher than that of the Beto-2 sample. should have been completed when the shearing reached
Figure 11.22 shows the simulation-test results on the steady state, the shear resistances at the steady state
sample Beto-4, taken near the toe of the deposits. The were almost the same as about 35 kPa. When the normal
apparent friction angle of Beto-4 was 5.0 degrees, the same stress became smaller from 620 kPa to 205 kPa, the ap-
Chapter 11 Experimental Study with Ring Shear Apparatus on the May 2004 LandslideDebris Flow at Bettou-dani Valley, Haku-san Mountain, Japan 165

1. Concentrated groundwater flows were a main trigger-

ing factor for the landslide initiation by increasing
water pressure in the slope;
2. In the ring-shear simulation test of the landslide initia-
tion, it was shown that even under naturally drained con-
ditions, the mobilized shear resistance of the weathered
soil in the source area showed a rapid decrease after land-
slide initiation, and this should be the instinctive factor
for rapid landslide motion after its initiation;
3. In the ring-shear simulation test of dynamic loading
on the valley deposits, it was shown that high poten-
tial for grain-crushing of upstream deposits and lower
potential of the downstream deposits controlled the
traveling and depositing process of the debris flow;
4. The shear resistance at steady state under undrained
conditions is the same for the soil samples taken from
different parts of the valley (Sample Beto-2, 3, 4). A
possible reason is that although the initial grain gra-
dations of these samples differ at the initial state, the
soil at the shear zone would become the same when
the shearing process reached the steady state, when all
of the possible grain-crushing is completed.


Deep thanks are given to the Kanazawa Office of Rivers

Fig. 11.22. Simulation test results on sample Beto-4 when debris flow and National Highways, MLIT, for cooperation in the field
traveled to and deposited near No. 10 debris retention dam.
a Applied-stress signals (normal stress and shear stress increments);
work and as a source of information on the May 2004 land-
b time-series data; c effective stress path (ESP) and total stress path slidedebris flow. Financial supports by research grants
(TSP). BD = 0.95 (No. 15310127, Representative: F.W. Wang) from the Min-
istry of Education, Culture, Sports, Science, and Technol-
parent friction angle became larger from 2.6 degrees to ogy of Japan (MEXT), and 16G-03 Joint Research Pro-
5.0 degrees; and when it became larger than the slope gramme of Disaster Prevention Research Institute (DPRI),
angle (it was about 5.0 degrees at the position near No. 10 Kyoto University are highly appreciated. Dr. Huabin Wang,
debris retention dam), the debris flow should have decel- Mr. Ryuta Saito, Mr. Jozef Jurko, and Mr. Taichi Minamitani
erated and finally came to a stop. of the Research Centre on Landslides (RCL), Disaster Pre-
From the above simulation tests that reproduced rapid vention Research Institute, Kyoto University, joined the
loading on valley deposits, the impact process, the trav- sampling and field investigation.
eling process, and depositing process of the debris flow
that occurred in May 2004 were well reproduced in the
laboratory. References
Kanazawa Office of Rivers and National Highways, Ministry of Land,
11.6 Conclusions Infrastructure and Transport of Japan (2004a) Investigation re-
port on Jinnosuke-dani landslide (in Japanese)
Kanazawa Office of Rivers and National Highways, Ministry of Land,
The May 2004 landslidedebris flow that occurred in the Infrastructure and Transport of Japan (2004b) Debris flow oc-
Bettou-dani of the Jinnosuke-dani landslide, Haku-san curred on 17 May 2004 in Bettou-dani. Newsletter Sabo Hakusan
Mountain, showed a fluidization process from landslide 6:14 (in Japanese)
to debris flow. By analysis of the monitored video images Kanazawa Office of Rivers and National Highways, Ministry of Land,
of the debris flow, field investigation of the source area of Infrastructure and Transport of Japan (2004c)
the landslide, and laboratory ring-shear tests that simu- Kaseno Y (1993) Geological mapping of Ishikawa. Geology Bulletin
lated the rainfall triggering mechanism and the fluidiza- (in Japanese)
tion mechanism during the process of downstream travel, Kaseno Y (2001) Supplement of geological mapping of Ishikawa.
the following were concluded: Geology Bulletin (in Japanese)
166 Fawu Wang Kyoji Sassa

Marsal RJ (1967) Large scale testing of rockfill materials. J Soil Mech Sassa K, Fukuoka H, Wang G, Ishikawa N (2004a) Undrained dy-
Found Div-ASCE 93(SM2):2743 namic-loading ring-shear apparatus and its application to land-
Okuno T, Wang FW, Matsumoto T (2004) The deforming char- slide dynamics. Landslides 1:719
acters of the Jinnosuke-dani landslide in Haku-san moun- Sassa K, Wang G, Fukuoka H, Wang FW, Ochiai T, Sugiyama M,
tainous area, Japan. In: Lacerda W, Ehrlich M, Fontoura S, Sekiguchi T (2004b) Landslide risk evaluation and hazard map-
Sayao A (eds) Landslides: Evaluation & Stabilization. Pro- ping for rapid and long-travel landslides in urban development
ceedings IX International Symposium on Landslides, Rio de area. Landslides 1(3):221235
Janeiro, 2, 12791285 Wang FW, Sassa K (2007) Initiation and traveling mechanisms of
Sassa K (1988) Geotechnical model for the motion of landslides. In: the May 2004 landslidedebris flow at Bettou-dani of the
Special Lecture of 5th International Symposium on Landslides, Jinnosuke-dani landslide, Haku-san Mountain, Japan. Soils and
Landslides, 1015 July, 1, pp 3755 Foundations 47(1) (in press)
Sassa K, Fukuoka H, Wang FW (1997) Mechanism and risk assess- Wang FW, Okuno T, Matsumoto T (2004) Deformation style and in-
ment of landslide-triggered-debris flows: lesson from the fluential factors of the giant Jinnosuke-dani landslide in Japan.
1996.12.6 Otari debris flow disaster, Nagano, Japan. In: Cruden In: Proceedings of the Fifteenth Southeast Asian Geotechnical
DM, Fell R (eds) Landslide risk assessment. Proceedings of the Conference, 1, pp 399404
International workshop on landslide risk assessment. Honolulu, Wang FW, Okuno T, Matsumoto T (2007) Deformation characteris-
1921 February, pp 347356 tics and influential factors for the giant Jinnosuke-dani landslide
Sassa K, Wang G, Fukuoka H (2003) Performing undrained shear in the Haku-san Mountain area, Japan. Landslides: Journal of the
tests on saturated sands in a new intelligent type of ring shear International Consortium on Landslides 4(1), DOI 10.1007/
apparatus. Geotech Test J 26(3):257265 s1034600600499 (in Press)
Chapter 12

On the Pore-pressure Generation and Movement

of Rainfall-induced Landslides in Laboratory Flume Tests

Gonghui Wang* Kyoji Sassa

Abstract. Using a small flume, a series of tests was conducted to trig- and are characterized by rapid movement and long runout
ger rainfall-induced landslides. Based on monitoring of sliding dis- distance. It is generally recognized that rainfall-induced
tance and pore pressures, the process of pore-pressure generation
landslides are caused by increased pore pressures and
in relation to sliding distance was examined. By performing tests
on sands of different grain sizes (silica sand no. 7 (D50 = 0.14 mm) seepage forces during periods of intense rainfall (Terzaghi
and no. 8 (D50 = 0.057 mm)) at different initial dry densities or dif- 1950; Sidle and Swanston 1982; Sitar et al. 1992; Ander-
ferent thicknesses, the effects of these factors on pore-pressure gen- son and Sitar 1995). It is the increased pore pressure that
eration and failure behavior of a landslide mass were analyzed. Re- decreases the effective stress in the soil and thus reduces
sults from tests of different initial densities showed that for each
the soil shear strength, eventually resulting in slope failure
sample there was an optimal density index, at which the pore pres-
sure build-up after failure reached its maximum value. This optimal (Brand 1981; Brenner et al. 1985). Further studies have il-
density index varied with the thickness of sample and the grain size lustrated that in some cases of rainfall-induced landslides,
of samples. Moreover, observed failure phenomena showed that movement along the sliding surface leads to crushing of
the failure mode also depended greatly on the grain size and sample the soil grains, which results in the liquefaction along this
thickness. In fact, flowslides were initiated in the tests on finer silica surface, finally resulting in rapid movement and long runout
sand (no. 8), whereas retrogressive sliding occurred in the tests on
silica sand no. 7. Results of tests on mixtures of silica sand no. 8 with
distance (Sassa 1996; Sassa 1998a, b); thus, high pore pres-
different contents of loess by weight showed that the existence of sure is a result of shearing along the sliding surface.
fine-particle soil (loess) could significantly change the flow behav- Liquefaction, a process by which the soil suddenly loses
ior of a landslide mass during motion: the flow behavior of soils with a large proportion of its shear resistance due to the gen-
20 and 30 percent loess was different from these two silica sands eration of high pore pressure, is always the reason for
and the mixture with 10 percent loess, showing greater velocity
fluidized landslides. Liquefaction triggered by dynamic
without deceleration. This suggests the existence of a mechanism
that maintains high pore pressures during motion for these soils. In effects, such as earthquakes, or by static effects, such as
addition, by rotating saturated samples in a double-cylinder appa- rainfall, snowmelt, etc., has been studied extensively (e.g.,
ratus, a mechanism was examined in which pore pressure in satu- Terzaghi 1956; Seed 1966, 1979; Bishop 1967, 1973; Castro
rated soils during motion was maintained. The results showed that 1969; Casagrande 1971; Castro and Poulos 1977; Sassa
the pore pressure of the saturated mixture increased with velocity
1984, 1996, 1998a, 1998b; Eckersley 1985, 1986; Hird and
because of the floating of sand grains that accompanied the move-
ment for each test. In addition, the sample with finer grain sizes or Hassona 1990; Ishihara et al. 1990; Ishihara 1993), and
greater fine-particle (loess) contents floated more easily, and high much knowledge has been accumulated about the mecha-
pore pressure could be maintained during motion. The floating ra- nism of liquefaction. However, most of the understand-
tios of grains reached a high value (>0.85) at a very slow velocity for ing is based on simple element testing of small specimens
samples with 20 and 30 percent loess. Based on these test results, it under idealized conditions (e.g., saturated, undrained, and
is concluded that grain size and fine-particle contents can have a
significant impact on the mobility of rainfall-induced landslides.